Cortical Circuits Underlying Social and Spatial Exploration in Rats

Cortical circuits underlying social and spatial exploration in rats

DISSERTATION zur Erlangung des akademischen Grades D o c t o r r e r u m n a t u r a l i u m (Dr. rer. nat.) im Fach Biologie eingereicht an der Lebenswissenschaftlichen Fakultät der Humboldt-Universität zu Berlin von M.Sc. Christian Laut Ebbesen

Präsidentin der Humboldt-Universität zu Berlin Prof. Dr.-Ing. Dr. Sabine Kunst

Dekan der Lebenswissenschaftlichen Fakultät Prof. Dr. Bernhard Grimm

Gutachter 1. Prof. Matthew Larkum, Ph.D.. Matthew Larkum, Ph.D. 2. Dr. James Poulet Dr . James Poulet 3. Prof. Dr. Michael Brecht Prof. Dr. Michael Brecht

Tag der mündlichen Prüfung: 6. Juli 2017

BERLIN SCHOOL OF MIND AND BRAIN berlin

Cortical circuits underlying social and spatial exploration in rats

Doctoral candidate: Christian Laut Ebbesen

Supervisors: Prof. Dr. Michael Brecht Prof. Dr. Gabriel Curio

April 2017

Berlin School of Mind and Brain Bernstein Center for Computational Neuroscience Animal Physiology / Systems Neurobiology and Neural Computation Humboldt Universität zu Berlin, Berlin, Germany

Ebbesen (2017) Table of Contents

1. Abstract 1 2. Zusammenfassung 3 3. General Introduction & Thesis Outline 4 3.1 General Introduction. 4 3.1.1 Why do we study the brain?...... 4 3.1.2 Three heuristics for investigating neural circuits...... 4 3.1.3 Neuroethology & functional localization ...... 5 3.1.4 How should we look a cortical data? . 7 3.2 Thesis Outline...... 8 3.2.1 Two investigations into cortical function ...... 8 3.2.2 Part 1: Parahippocampal cortex and neural circuits underlying spatial exploration.9 3.2.3 Part 2: Motor cortex and neural circuits underlying social exploration . 12 3.3 References...... 15 4. Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex 22 4.1 Introduction...... 23 4.2 Results...... 23 4.3 Discussion...... 28 4.4 Experimental Procedures . 28 4.5 References...... 29 4.6 Supplemental Information...... 30 5. Functional Architecture of the Rat Parasubiculum 52 5.1 Introduction...... 53 5.2 Materials and Methods. 54 5.3 Results...... 56 5.4 Discussion...... 59 5.5 References...... 64 6. Cell Type-Specific Differences in Spike Timing and Spike Shape in the Rat Parasubiculum and Superficial Medial Entorhinal Cortex 66 6.1 Introduction...... 68 6.2 Results...... 69 6.3 Discussion...... 74 6.4 Experimental Procedures . 76 6.5 References...... 77

I Ebbesen (2017)

6.6 Supplemental Information...... 79 7. Vibrissa motor cortex activity suppresses contralateral whisking behavior 90 7.1 Introduction...... 93 7.2 Results...... 94 7.3 Discussion...... 99 7.4 References...... 102 7.5 Figures...... 107 7.6 Methods...... 120 7.7 Methods References...... 128 8. Motor cortex – To act or not to act? 130 8.1 Introduction...... 133 8.2 From motor cortex to muscle output. 135 8.3 Inactivation studies: Acting without motor cortex...... 142 8.4 Motor cortex – not to act?...... 144 8.5 Conclusions...... 147 8.6 References...... 150 8.7 Figures...... 161 9. General Discussion 168 9.1 Parahippocampal cortex and neural circuits underlying spatial exploration. 168 9.1.1 Structure-function relationships in parahippocampal cortex . 168 9.1.2 How are grid cells made?. 168 9.1.3 What does the grid cell system teach us about cortical computation? . . . . 170 9.2 Motor cortex and neural circuits underlying social exploration. 172 9.2.1 What is the function of rat vibrissa motor cortex?...... 172 9.2.2 Social computations in sensorimotor cortex...... 173 9.3 References...... 174 10. Curriculum vitae 182 11. List of publications 184 12. Declaration of contribution 185 13. Acknowledgements 190

PLACER

II Ebbesen (2017) List of Figures

Figure 1 The cerebral cortex is the outermost cell layers of the mammalian telencephalon.. . . 5 Figure 2 An example of viewing cortical data in the ‘right’ way? ...... 7 Figure 3 Schematic showing the location of motor cortex and parahippocampal cortex in the rat brain...... 9 Figure 4 Hippocampal place cells. . 9 Figure 5 Spatial discharge patterns of neurons in parahippocampal cortex...... 10 Figure 6 Social facial touch...... 13 Figure 7 Movement suppression is under-represented in depictions of motor cortex . . . . 14 Figure 8 Graphical abstract outlining the main message of this chapter. . 22 Figure 9 A Wistar rat it reins. . 90 Figure 10 Modular cytoachitectonics of parahippocampal cortex...... 148 Figure 11 Grid-cell-like responses of entorhinal neurons to saccades across a visual scene. . . 149 Figure 12 What are cortical neurons doing? ...... 151

III Ebbesen (2017)

IV Ebbesen (2017) Declaration I declare that the doctoral thesis entitled Cortical circuits underlying social and spatial exploration in rats represents an original work of the author apart from the references and declared contributions under the provisions § 6 (3) of the doctoral degree regulations, dated 5 March 2015, of the faculty of Life Sciences of Humboldt-Universität zu Berlin. The work involved no collaborations with com- mercial doctoral degree supervisors. I affirm that I have neither applied for nor hold a corresponding doctoral degree and this work has not been submitted in full or part to another academic institution. Further, I acknowledge the doctoral degree regulations which underlie this procedure, and state that I abided by the principles of good academic practice of Humboldt-Universität zu Berlin.

Christian Laut Ebbesen, April 2017

V Ebbesen (2017)

VI Ebbesen (2017) Abstract 1. Abstract In order to understand how the mammalian brain works, we must investigate how neural activity contributes to cognition and generates complex behavioral output. In this thesis I present work, which focuses on two regions of the cerebral cortex of rats: parahippocampal cortex and motor cortex. In the first part of the thesis we investigate neural circuits in the parasubiculum and the superficial medial enthorhinal cortex, two structures that play a key role in spatial cognition. Briefly, we find that the in these regions, anatomical identity and microcircuit embedding is a major determinant of both spatial discharge patterns (such as the discharge patterns of grid cells, border cells and head-direction cells) and temporal coding features (such as spike bursts, theta-modulation and phase precession). In the second part of the thesis we investigate the activity of neurons in vibrissa motor cortex during complex motor behaviors, which play a vital role in rat ecology: self-initiated bouts of exploratory whisking in air, whisking to touch conspecifics during social interactions and whisking to palpate objects. Briefly, we find that neural activity decreases during whisking behaviors, that microstimulation leads to whisker retraction and that pharmacological blockade increases whisker movement. Thus, our observations collectively suggest that a primary role of vibrissa motor cortex activity is to suppress whisking behaviors. The second part of the thesis concludes with a literature review of motor suppressive effects of motor cortical activity across rodents, primates and humans to put this unexpected finding in a broader context.

1 Ebbesen (2017)

2 Ebbesen (2017) Zusammenfassung 2. Zusammenfassung Um zu verstehen, wie das Gehirn von Säugetieren funktioniert, untersuchen wir wie neuronale Aktivität einerseits zu Kognition beträgt und andererseits komplexe Verhaltensweisen ermöglicht. Im Fokus dieser Doktorarbeit stehen dabei zwei Regionen der Großhirnrinde der Ratte: der para- hippocampale Cortex und der motorische Cortex. Im ersten Teil haben wir neuronale Schaltkreise im parahippocampalen Cortex und in den oberen Schichten des enthorhinalen Cortex untersucht, während Ratten ihre Umgebung räumlich erkunden. Diese beiden Regionen tragen wesentlich zum Orientierungssinn bei. Dabei haben wir herausgefunden, dass anatomische Identität und Einbindung in den Microschaltkreis einerseits räumliche neuronale Signale, wie zum Beispiel der Aktivität von grid cells, border cells und head-direction cells, bestimmen. Andererseits tragen diese beiden Eigen- schaften auch zur temporalen Präzision neuronaler Signale bei, wie zum Beispiel in Form von spike bursts, theta Modulation und phase precession. Im zweiten Teil dieser Doktorarbeit untersuchen wir die Aktivität von Neuronen im Vibrissen Motorcortex während komplexer Bewegungsabläufe der Schnurrhaare, die dem natürlichen Repertoire der Ratte entstammen: eigeninitiierte Bewegungen in freier Luft, Berührung von Artgenossen zur sozialen Interaktion und das Abtasten von Objekten. Da- bei haben wir herausgefunden, dass neuronale Aktivität im Motorcortex während der Bewegung der Schnurrhaare unterdrückt ist, dass elektrische Microstimulation zum Rückzug der Schnurrhaare führt und, dass pharmakologische Blockade Bewegung der Schnurrhaare fördert. Um diese überraschende Beobachtung in einen breiteren Kontext zu integrieren, endet dieser Teil mit einer Bewertung der Literatur zu der bewegungsunterdrückenden Wirkung von Motorcortex Aktivität bei Nagetieren, Pri- maten und Menschen.

3 Ebbesen (2017) Chapter 2 3. General Introduction & Thesis Outline 3.1 General Introduction 3.1.1 Why do we study the brain? Three major factors make the mammalian forebrain one of the most interesting structures in the universe: a perspective of basic health care, a perspective of introspection and a political perspective. First, a better understanding of the mammalian neurobiology would be immensely helpful in designing novel therapeutic strategies for mental conditions. These conditions have high incidence and major impact on life quality, but presently totally dissatisfactory treatment options (Connell et al., 2014; Diener et al., 1999; Lehman, 1996; WHO, 1995). Thus, advances in this regard would have massive positive effects on public welfare. Secondly, understanding the neural circuits, which give rise to our own conscious first-person perspective, is intrinsically interesting. It provides a unique avenue to explore our own subjectivity as humans and as part of the mammalian family. Thirdly, while some aspects of political ideology are strictly normative, they also often encompass (in a more or less explicit manner) assumptions about neurobiology. Suppositions about how e.g. reward, competition and motivation modulate behavior are ultimately statements about brain function, which presently remain largely unverified. Clearly, any extrapolation from scientific findings to societal policy must be extremely cautious (cf. historical ‘scientific’ racism, classism and heteronormativity (Bashford and Levine, 2010; Belkhir, 1994)). Nevertheless, aligning ‘folk psychology’ assumptions underlying how e.g. society and economics is currently structured more closely with empirical studies of the neural control of behavior is bound to positively influence how we collectively organize (Churchland, 1981; Fitzgerald et al., 2014; Žižek, 1989).

3.1.2 Three heuristics for investigating neural circuits To effectively investigate how brain structure contributes to the generation of cognition and complex behavioral output, we must design our studies such that three practical conditions are satisfied: First, we must be able to record and manipulate neural activity with high fidelity and high temporal and anatomical resolution. Secondly, we must investigate neural activity when the animal is doing behaviors, which actually drive the neural circuits in a way, which allows them to reveal their function experimentally. Finally, when we have good data from interesting behaviors, we must analyze and look at these data in an appropriate way.

In the last decade, our capability in the first domain has made immense advances. There has been

4 Ebbesen (2017) Introduction & Outline a massive progress in the technical tools, which we can use to monitor and manipulate neural activity of awake, behaving animals with (sub)-cellular resolution, e.g. abundantly available transgenic animals and viral tools (Harris et al., 2014; Heldt and Ressler, 2009; Witten et al., 2011), optogenetics (Deisseroth, 2015; Kim et al., 2017), DREADDs (Whissell et al., 2016), in-vivo multi-photon imaging of various sensors (Broussard et al., 2014; Yang and Yuste, 2017), high-density electrophysiology (Buzsáki et al., 2015), advanced statistical/ data-mining methods (Aljadeff et al., 2016; Harris et al., 2016), etc. However, our main challenge is not only to record neural activity from an ever-increasing number of cells with ever-increasing fidelity. Rather, we also need advances in our ability to satisfy the second and third condition, i.e. to identify correct behaviors to drive cortical activity and to analyze and view the data in the proper way. This is beautifully illustrated by two classic studies on the mammalian cerebral cortex: The discovery of motor cortex by Fritch and Hitzig (Fritsch and Hitzig, 1870) and the discovery of orientation tuning in visual cortex by Hubel and Wiesel (Hubel and Wiesel, 1959).

3.1.3 Neuroethology & functional localization The cerebral cortex (Figure 1) is the most recently evolved brain structure across phylogeny and a massive amount of evidence coherently points to a major role of cortex Figure 1: The cerebral cortex is the out- in ‘high-level’ cognitive capacities (DeFelipe, 2011; Mol- ermost cell layers of the mammalian nár et al., 2014). The neural ‘workload’ is not uniformly telencephalon. Top to bottom: Brain of human, lion, grey kangaroo and polar bear distributed across cortical neurons, but rather, certain (coronal sections, thionin stain, cell bodies domains of neural computation are – at least to some appear dark purple in color). Scale bar = 1 cm. Adapted with permission from http:// degree – localized to distinct cortical regions (Brett et www.brains.rad.msu.edu, supported by the US National Science Foundation. 5 Ebbesen (2017) Chapter 2 al., 2002; Campbell, 1905; Zola-Morgan, 1995). This cortical localization can be revealed by inter- ventions, which manipulate neural activity in a pinpointed way, for example by the focal inactivation (classically e.g. by lesioning) or activation (classically e.g. by intra-cortical electrical stimulation) of neural activity. Such localized perturbations of cortical activity lead to highly domain-specific changes (i.e. deficits) in cognitive or behavioral function.

The first and most ‘classic’ cortical localization experiment was done in Berlin in the 1870s. Two young doctors discovered, that when they applied weak stimulation currents to the cortical surface of anaesthetized rabbits and dogs, they would sometimes elicit muscle twitches. These twitches were only elicited in certain, specific parts of the frontal cortex, a brain region we now refer to as primary motor cortex (Fritsch and Hitzig, 1870). This discovery was a major breakthrough in cortical physiology, since it gave a hint that neurons in this part of the cortex must be somehow involved the control of muscle output, i.e. motor control. Thus, to investigate these function of these neurons, subsequent investigations in awake animals and humans have focused on recording the activity patterns of motor cortical neurons during various behaviors, which require movement, muscle action and fine motor control (Graziano, 2011; Lemon, 2008; Shenoy et al., 2013).

Such studies on motor cortical activity during motor behavior have yielded major insights (Lemon, 2008; Shenoy et al., 2013), and exemplify the power of recording the activity pattern of neurons during behaviors, which actually drive their activity in a natural, biological way. We can draw a general conclusion: in order to satisfy the second heuristic mentioned above, neurobiology must rely heavily on ethological approaches. In order to investigate neural coding, we must incorporate findings from the study of animal behavior in natural conditions and design our experiments such that we investigate neural signals in animals performing behaviors, which resemble the naturally occurring environmental challenges that these neural circuits have evolved to overcome (e.g. the activity of motor cortical neurons during motor behaviors) (Krakauer et al., 2017). If we design our experiments such that the experimental animal must e.g. repeatedly perform an experimental task (e.g. respond to a repeated sensory stimulus (Connor et al., 2010; O’Connor et al., 2009)), we are almost certain to find neurons which spuriously correlate with task-related parameters, simply due to the repeated structure of the task design. The neural activity pattern may not in any meaningful way reflect what these neurons are ‘really doing’ (Krakauer et al., 2017).

6 Ebbesen (2017) Introduction & Outline

3.1.4 How should we look a cortical data? a Stimulus orientation Stimulus on Neural signals are recorded in the temporal domain, so relating the function of cortical spikes to behavior requires appropriate mathematical transforma- tions. This means, that even if we can record neural activity with high fidelity during relevant, ecological behavior, where these neurons are highly involved in the task, it is not guaranteed that we will gain any new insights about neural function. In fact, major breakthroughs in our understanding of neural coding have not been driven by technical advances, but by conceptual advances: By looking at neural data in the ‘right’ way.

The Nobel-Prize winning discovery of visual orientation tuning by Hubel and Wiesel is an example of this principle. At the time of their discovery, anatomical observations and observations on behavioral changes following cortical lesions suggested b that striate cortex (now ‘primary visual cortex’) was involved in visual processing (Hubel and Wiesel,

1998; Wurtz, 2009). However, it was unclear how Spike rate neurons physically implemented the representation of visual stimuli. It was known that the neuronal response to a circular light stimulus, a classically used Stimulus orientation Figure 2: An example of viewing cortical data visual stimulus at the time, depended on the retinal in the ‘right’ way? (a) Example responses (right) location. Some retinal locations elicited an increase of a single neuron in cat visual cortex to various orientations of light bar stimulus (left). Sin- in spike rate, and some retinal locations lead to sup- gle spikes are visible as vertical lines, stimulus pression of spike rates, but there were no obvious presened for 1s. (b) Plotting the spike rate during stimulus presentation as a function of the correlations between the stimulus and the spike dis- stimulus orientation reveals a clear and elegant charges (Wurtz, 2009). Hubel & Wiesel discovered pattern: the neuron is symmetrically tuned for a specific orientaion (schematic depicts a von that if they recorded the activity of single neurons Mises pdf). (Adapted from Hubel & Wiesel, when the cat was being presented with a light bar 1959, with permission, Wiley and Sons) 7 Ebbesen (2017) Chapter 2 stimulus at various orientations (Figure 2a) and the neuronal discharge rate was plotted as a function of the stimulus orientation, a simple and elegant pattern emerged: single neurons showed sharp, symmetric tuning to the orientation of the bar (Figure 2b) (Hubel and Wiesel, 1959).

Discovering such a beautiful pattern makes it appear likely, that we are now looking at the neural data in the ‘correct’ way. However, it is important to keep in mind, that from a standpoint of information theory, there is in principle no reason why neurons should implement neural algorithms in a way such that the activity of single cells display beautiful patterns. In a wide range of computational problems, it is fully possible for a network to reach equal (and better) performance using distributed network representations, which have no obvious patterns or ‘aesthetic’ qualities when considering intermedi- ate levels or subsets of nodes (Borst and Theunissen, 1999; Dayan et al., 2011; Fairhall et al., 2012; Harris, 2005; Kaufman and Churchland, 2016; Shenoy et al., 2013). A similar argument can be made about physical laws in general. There is – in principle – no reason why it should be possible to express fundamental laws of e.g. field theory or quantum mechanics in simple, elegant equations. Nonetheless, it is possible and seems to be a deep organizational principle of known physics (Elliott and Dawber, 1979; Feynman et al., 1963).

The concept of searching for elegant patterns in nature has historically been an incredible successful guiding heuristic in physics. In physics, there are statistical arguments which may provide some expla- nations as to why the patterns we observe in nature are so ‘neat’ (Hardy, 2001; Aaronson, 2004), but these arguments are not easy to make about biological systems which have arisen through evolution. Nonetheless, in cortical physiology, we may use the discovery of easily interpretable spiking patterns (e.g. clear orientation tuning), as a heuristic principle to guide investigations into neural coding in the cortex: If we discover neurons with obvious, elegant patters, this can serve as a good starting point for our investigations.

3.2 Thesis Outline 3.2.1 Two investigations into cortical function In this thesis, I will present work, which focuses on two parts of the cerebral cortex of rats: the parahippocampal cortex and motor cortex (Figure 3). We have investigated the neural activity in these two cortical regions during two kinds of complex, ecological behaviors: spatial exploration and social exploration. Our knowledge about the two cortical regions, which we have studied, aligns differentially

8 Ebbesen (2017) Introduction & Outline with the three heuristics discussed above. Conse- Motor cortex Hippocampus Parahippocampal cortex quently, the questions we could ask and the scientific contributions we could make to the study of these two cortical areas have different characteristics. In this thesis, I will only present work from studies during PhD, where I had a major contribution (i.e. first Figure 3: Schematic show- or co-first authorships). For a full list of publications during my ing the location of motor cortex (red) and parahippocampal PhD studies, please refer to Chapter 11. cortex (purple) in the rat brain. (Adapted from Poucet & Sargolini, 3.2.2 Part 1: Parahippocampal cortex and neural circuits 2013, w. permission, Nature NPG) underlying spatial exploration In the first part of the thesis, we investigate the activity in parahippocampal cortex during spatial exploration. Specifically, we look at activity in the superficial layers of the medial entorhinal cortex and at the parasubiculum, when rats are foraging for chocolate rewards in an open arena, in darkness.

Even though such a chocolate foraging task appears very simple, it forces the animal to perform sophisticated neural computations. The fact that the task is done in darkness means that the animal is excluded from using distant visual cues to guide its movement. Rather, to navigate the open arena, the animal must integrate haptic, olfactory and proprioceptive information to generate an internal cognitive representation of the external space. Famously, in the dorsal hippocampus, there is a population of neurons which discharge spikes only when the animal is moving in specific parts of space (‘place Figure 4: Hippocampal place cells. Top: setup for record- cells’, O’Keefe and Dostrovsky, 1971; Muller et al., 1987; Figure 4). ing behavior (‘headlights’ and Numerous lesion studies point to a major role of these neurons in the camera) and neural activity (to commutator, etc.) of a freely generation of an internal cognitive ‘map’ (Bird and Burgess, 2008; moving rat. Below: Firing Eichenbaum, 2017; McNaughton et al., 2006; Moser et al., 2008). probabilty of a hippocampal neuron as a function of the rat’s location in the arena (yel- The medial entorhinal cortex is an elongated structure at the most low = low, dark blue = high probability). (Adapted from posterior end of the rodent cerebral cortex, which provides a massive, Muller et al., 1987, with permission, Soc. for Neurosci) 9 Ebbesen (2017) Chapter 2 excitatory input to the hippocam- Grid Border Head-direction pus (Ray et al., 2014; Varga et al., 2010). Since the medial entorhinal cortex is ‘upstream’ of the hippocampal place cells, several studies have investigated the activity patterns in medial entorhinal cortex during spatial behavior to elucidate which kind of ‘spatial’ input patterns hippocampal neurons might inherit from the Figure 5: Spatial discharge patterns of neurons in parahippocampal cortex: A grid cell, a border cell and a head-direction cell. Top medial entorhinal cortex (Bush et row shows running trajectory (black line) and spatial locations of al., 2014; Rowland et al., 2016). spikes of a single neuron (red dots) from a rat exploring a square arena, or a sqare arena with a high ‘wall-like’ obstacle in the These studies have described an center. Bottom row shows the spiking probability as a function array of striking spatial firing pat- of the rat’s location in the arena (blue = low, red = high prob.) or as a function of the rat’s heading direction (indicated by compass terns, most prominently grid cells cross). (Adapted from Hartley et al. 2014, with permission, The (neurons, which discharge in a Royal Soc.). regular, hexagonal ‘grid’ pattern, Figure 5, left, Hafting et al., 2005), border cells (neurons, which discharge when the animal is near borders in the environment, Figure 5, middle, Solstad et al., 2008) , band cells (neurons which discharge in periodic ‘bands’ across the arena, Krupic et al., 2012) and head-direction cells (neurons, which discharge when the animal is facing a specific direction, Figure 5, right, Taube et al., 1990; Boccara et al., 2010).

When I started my PhD work, there was already a wealth of theoretical and computational models, which proposed various cellular-level and network-level mechanisms, which could generate hippocampal place cells from spatial inputs from the entorhinal cortex (Burgess and O’Keefe, 2011; Giocomo et al., 2011; Zilli, 2012). Layer 2 of the medial entorhinal cortex contains two excitatory cell types: stellate neurons and pyramidal neurons (Alonso and Klink, 1993) (arguably with additional subdivisions, Fuchs et al., 2016). These two cell types are arranged in a remarkable modular fashion, which mirrors the hexagonal pattern of the grid cells (Kitamura et al., 2014; Ray et al., 2014). Crucially, these two cell types have very different projection patterns: stellate neurons project primarily to the dentate gyrus (Ray et al., 2014; Varga et al., 2010), while pyramidal send a comparatively much smaller projection to CA1 (Kitamura et al., 2014). Thus, the interpretation of what e.g. the entorhinal grid cells contribute 10 Ebbesen (2017) Introduction & Outline to spatial coding in the hippocampus depends on which hippocampal subfields receive these inputs. In Chapter 4 of this thesis, I present a study, where we used statistical machine learning methods combined with juxtacellular recordings from identified neurons and tetrode recordings from unidentified neurons to investigate how spatial response patterns (grid, border, head-direction) map onto stellate and pyramidal neurons:

(*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Spicher, D. & Brecht, M. (2014) Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex. Neuron, 84(6):1191-1197.

The parasubiculum is a thin structure, which wraps around the medio-dorsal edge of the medial entorhinal cortex (Andersen et al., 1971), receives various cortical and subcortical input and sends a major projection to the superficial layers of medial entorhinal cortex (Caballero-Bleda and Witter, 1993, 1994; van Groen and Wyss, 1990). In Chapter 5 of this thesis I present a study, where we investigate which kind of spatial and temporal response patterns are already present in the parasubiculum (Boc- cara et al., 2010), ‘upstream’ of the medial entorhinal cortex, and how the modular cytoarchitechture of the medial entorhinal cortex determines how entorhinal microcircuits may inherit such information from the parasubiculum:

The role of hippocampal spike timing in (spatial) memory is the most studied example of temporal coding in all of neuroscience (Colgin, 2016; Hasselmo, 2005; Howard and Eichenbaum, 2015). Despite the enormous scientific interest, we still know surprisingly little about how temporal coding features like spike bursts, theta-modulation (rhythmicity, locking, skipping) and phase precession map onto hippocampal and parahippocampal microcircuits. The paucity of data on the relationship between phase precession and microcircuits reflects the fact that the majority of studies have recorded tetrode data. This has given rise to a pleathora of theory and modeling of how the temporal coding patterns are generated and what their function might be (Colgin, 2016; Giocomo et al., 2011; Hasselmo, 2005; Zilli, 2012). The only way to prune this forest of models is to establish how temporal coding maps onto anatomically distinct cell types and microcircuits. In Chapter 6 of this thesis, I present a 11 Ebbesen (2017) Chapter 2 study where we analyze a large sample of cells recorded juxtacellularly in freely moving rats and ask how temporal coding maps onto the modular organization of parahippocampal cortex:

Ebbesen, C.L., Reifenstein, E.T., Tang, Q., Burgalossi, A., Ray, S., Schreiber, S., Kempter, R. & Brecht, M. (2016) Cell type-specific differences in spike timing and spike shape in rat parasubiculum and superficial medial entorhinal cortex.Cell Reports 16(4):1005-1015.

3.2.3 Part 2: Motor cortex and neural circuits underlying social exploration When we studied spatial coding in the medial entorhinal cortex, we could align our intuitions with the third heuristic mentioned above: the remarkable spatial response patterns of e.g. grid cells and border cells strongly suggests that neurons in this cortical area are somehow involved in the mental representation of allocentic space. However, perturbation of the activity in medial entorhinal cortex only has very subtle effects on behavior, and the ‘function’ of the grid cell systems still remains unresolved (Van Cauter et al., 2013; Hales et al., 2014; Sasaki et al., 2014). In contrast, our studies of the vibrissa motor cortex align much more closely with the second heuristic mentioned above: From anatomical studies and microstimulation experiments, there is strong evidence that neurons in vibrissa motor cortex are highly involved in the muscular control of whisking output (Brecht et al., 2004; Gioanni and Lamarche, 1985; Hall and Lindholm, 1974; Neafsey et al., 1986). However, while we have good indications of the ‘function’ of these neurons, we know of no way of plotting the activity of motor cortical neurons in a way, which generates as elegant and striking patterns as the ‘spatial’ responses found in entorhinal cortex. Previous investigations had identified a subpopulation of neurons in vibrissa motor cortex, which show significant correlations with whisking kinematics, such as whisking amplitude or phase of the whisking cycle, but these neurons are rare and the correlations are generally weak (Carvell et al., 1996; Friedman et al., 2012; Gerdjikov et al., 2013; Hill et al., 2011, Kleinfeld et al., 1999).

It remains an open question whether the striking patterns of e.g. spatial responses in entorhinal cortex is a general feature of cortical spike trains. We do not know whether a mathematical transformation exists, such that also e.g. motor cortical neurons would display spike patters during some motor behaviors, which are as easily interpretable as grid cells. Perhaps no such transformation exists (as is predicted by theories interpreting motor cortical activity patterns as trajectories in a high-dimensional dynamical system, Shenoy et al., 2013), or perhaps the highly controlled and ‘reduced’ motor behaviors of most previous investigations into activity patterns in vibrissa motor cortex (generally head-fixed ani-

12 Ebbesen (2017) Introduction & Outline mals whisking in air) does not drive the cortical network in a way where obvious patterns reveal themselves (Krakauer et al., 2017). In Chapter 7 of this thesis, I present a study where we contributed to resolve this question by investigating the activity of neurons in vibrissa motor cortex in freely moving animals during natu- Figure 6: Social facial touch is a delicate, multisensory be- ralistic behaviors, which are part of havior, where rats put their noses together and palpate each other’s faces with their whiskers, while they emit ultrasound their ecological repertoire of whisker vocalizations and sniff each others breath and pheromones. behaviors (Deschênes et al., 2012; The whisking patterns during such social interactions are complex and require fine motor control.The picture is record- Grant et al., 2012; Sachdev et al., ed with a highspeed camera under infrared light (i.e. in visual 2002; Welker, 1964; Wolfe et al., darkness for the rats), overlayed colors indicate sex of the rats. 2011): self-initiated bouts of exploratory whisking in air, whisking to touch conspecifics during social interactions (Figure 6) and whisking to palpate objects.

Ebbesen, C.L., Doron, G., Lenschow, C., Brecht, M. (2017). Vibrissa motor cortex activity suppresses contralateral whisking behavior. Nature Neuroscience 20(1):82-89

In the above study, we found that allowing the animals to freely perform more complex movement patterns allowed us to identify overall patterns in the modulation of motor cortical activity during whisker behaviors. However, surprisingly, we most often observed that motor cortical activity decreased during whisking behavior. Ever since the discovery of motor cortex about a 150 years ago (Fritsch and Hitzig, 1870), the prime function attributed to this cortical region has been generation of movement, hence the name „motor’ cortex (Lemon, 2008). However, our observations on vibrissa motor cortex (across recording, microstimulation and inactivation) collectively point to the conclusion that the primary role of vibrissa motor cortex activity is to suppresses whisker movements (i.e. this cortical area serves a ”brake“ rather than ”motor” function).

When we presented our work at conferences, the response of our colleagues was also often surprise, since it was generally assumed that rodent motor cortex should respond to movement with an increase of activity, as is classically the case in e.g. monkey distal limb motor cortex. However, our observations 13 Ebbesen (2017) Chapter 2

Figure 7: Movement suppression is under-represented in depictions of motor cortex (a) A textbook illustration of the human ‘motor’ (red) and somatosensory (green) cortex after the intraoperative stimulation experiments by Penfield & Rasmussen. The textbooks interpret human motor cortex as a ‘motor homunculus’, that is a muclelotopic motor map: “Motor cortex: movement” (red underscore, added) (Adapted from Heeger, 2006, with permission). (b) Intraoperative photograph of an actual human motor (red, color added) and somatosensory cortex (green, color added) mapping experiment, as reported in the famous book by Penfield & Rasmussen (1952). Right, we see what the patient reports, when site 13 and 14 (yellow arrow, middle of motor cortex) are stimulated: a clear suppression of movement. (Adapdted from Pen- field and Rasmussen, 1952, with permission) 14 Ebbesen (2017) Introduction & Outline are actually not unique, and several recent investigations of the function of motor cortices (especially in rodents) have found a puzzling predominance of neurons, which decrease their firing with movement and large amounts of movement-related motor cortical inhibition. However, while many authors have made observations, which indicate that motor cortex sometimes exerts a ‘negative control’ of motor output, there observations generally receive little attention. For example, Figure 7a shows a schematic included in almost every neuroscience textbook: a depiction of human motor cortex with a ‘motor homunculus’, which symbolizes a muclelotopic motor map. What is very little known and greatly underrated however, is that almost all stimulation effects observed by Penfield and Rasmussen were motor suppressive, to quote from the original paper: ‘Oh, my right hand. I couldn’t move it.’ (Pen- field and Rasmussen, 1952). In Chapter 8 of this thesis, I conclude by presenting a review highlighting both ‘classic’ and recent observations on the primary motor cortex, which point to major movement- suppressive functions. As illustrated with the example of the textbook-figure above, there is a tendency that motor cortex findings are presented in a way, which is mostly centered around movement generation, and there is a tradition of attaching less significance to motor suppressive effects. We therefore think that a coherent presentation of motor suppressive effects could have a positive impact on the field and contribute to a more balanced view of motor cortex function by emphasizing movement- suppression as a neglected theme in motor control:

Ebbesen, C.L., Brecht, M. (2017) Motor cortex – to act or not to act? (in review at Nature Reviews Neuroscience)

3.3 References Aljadeff, J., Lansdell, B.J., Fairhall, A.L., Kleinfeld, D., Adelman, T.L., Bialek, W., Olberg, R.M., Arcas, B.A. y, Fairhall, A.L., Akaike, H., et al. (2016). Analysis of Neuronal Spike Trains, De- constructed. Neuron 91, 221–259. Alonso, A., and Klink, R. (1993). Differential electroresponsiveness of stellate and pyramidal-like cells of medial entorhinal cortex layer II. J. Neurophysiol. 70, 128–143. Andersen, P., Bliss, T.V.P., and Skrede, K.K. (1971). Lamellar organization of hippocampal excitatory pathways. Exp. Brain Res. 13, 222–238. Bashford, A., and Levine, P. (2010). The Oxford Handbook of the History of Eugenics (Oxford: Ox- ford University Press). Belkhir, J. (1994). Race, Sex, Class & “Intelligence” Scientific Racism, Sexism & Classism. Race, Sex

15 Ebbesen (2017) Chapter 2

Cl. 1, 53–83. Bird, C.M., and Burgess, N. (2008). The hippocampus and memory: insights from spatial processing. Nat. Rev. Neurosci. 9, 182–194. Boccara, C.N., Sargolini, F., Thoresen, V.H., Solstad, T., Witter, M.P., Moser, E.I., and Moser, M.-B. (2010). Grid cells in pre- and parasubiculum. Nat. Neurosci. 13, 987–994. Borst, A., and Theunissen, F.E. (1999). Information theory and neural coding. Nat. Neurosci. 2, 947–957. Brecht, M., Krauss, A., Muhammad, S., Sinai-Esfahani, L., Bellanca, S., and Margrie, T.W. (2004). Organization of rat vibrissa motor cortex and adjacent areas according to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells. J. Comp. Neurol. 479, 360–373. Brett, M., Johnsrude, I.S., and Owen, A.M. (2002). The problem of functional localization in the human brain. Nat. Rev. Neurosci. 3, 243–249. Broussard, G.J., Liang, R., and Tian, L. (2014). Monitoring activity in neural circuits with genetically encoded indicators. Front. Mol. Neurosci. 7, 97. Burgess, N., and O’Keefe, J. (2011). Models of place and grid cell firing and theta rhythmicity. Curr. Opin. Neurobiol. 21, 734–744. Bush, D., Barry, C., and Burgess, N. (2014). What do grid cells contribute to place cell firing? Trends Neurosci. 37, 136–145. Buzsáki, G., Stark, E., Berényi, A., Khodagholy, D., Kipke, D.R., Yoon, E., and Wise, K.D. (2015). Tools for Probing Local Circuits: High-Density Silicon Probes Combined with Optogenetics. Neuron 86, 92–105. Caballero-Bleda, M., and Witter, M.P. (1993). Regional and laminar organization of projections from the presubiculum and parasubiculum to the entorhinal cortex: An anterograde tracing study in the rat. J. Comp. Neurol. 328, 115–129. Caballero-Bleda, M., and Witter, M.P. (1994). Projections from the presubiculum and the parasubiculum to morphologically characterized entorhinal-hippocampal projection neurons in the rat. Exp. Brain Res. 101, 93–108. Campbell, A.W. (1905). Histological studies on the localization of cerebral function. (Cambridge: The Macmillan Company). Carvell, G.E., Miller, S.A., and Simons, D.J. (1996). The relationship of vibrissal motor cortex unit activity to whisking in the awake rat. Somatosens. Mot. Res. 13, 115–127. Van Cauter, T., Camon, J., Alvernhe, A., Elduayen, C., Sargolini, F., and Save, E. (2013). Distinct Roles of Medial and Lateral Entorhinal Cortex in Spatial Cognition. Cereb. Cortex 23, 451– 16 Ebbesen (2017) Introduction & Outline

459. Churchland, P. (1981). Eliminative materialism and the propositional attitudes. J. Philos. 78, 67–90. Colgin, L.L. (2016). Rhythms of the hippocampal network. Nat. Rev. Neurosci. 17, 239–249. Connell, J., O’Cathain, A., and Brazier, J. (2014). Measuring quality of life in mental health: are we asking the right questions? Soc. Sci. Med. 120, 12–20. Connor, D.H.O., Clack, N.G., Huber, D., Komiyama, T., Myers, E.W., and Svoboda, K. (2010). Vibrissa-Based Object Localization in Head-Fixed Mice. 30, 1947–1967. Dayan, P., Feller, M., and Feldman, D. (2011). Networks, circuits and computation. Curr. Opin. Neurobiol. 21, 661–663. DeFelipe, J. (2011). The evolution of the brain , the human nature of cortical circuits , and intellectual creativity. 5, 1–17. Deisseroth, K. (2015). Optogenetics: 10 years of microbial opsins in neuroscience. Nat. Neurosci. 18, 1213–1225. Deschênes, M., Moore, J., and Kleinfeld, D. (2012). Sniffing and whisking in rodents. Curr. Opin. Neurobiol. 22, 243–250. Diener, E., Suh, E.M., Lucas, R.E., and Smith, H.L. (1999). Subjective well-being: Three decades of progress. Psychol. Bull. 125, 276–302. Eichenbaum, H. (2017). The role of the hippocampus in navigation is memory. J. Neurophysiol. 117, 1785–1796. Elliott, J.P., and Dawber, P.G. (1979). Symmetry in physics (Oxford: Macmillan Press). Fairhall, A., Shea-Brown, E., and Barreiro, A. (2012). Information theoretic approaches to understanding circuit function. Curr. Opin. Neurobiol. 22, 653–659. Feynman, R.P. (Richard P., Leighton, R.B., and Sands, M.L. (Matthew L. (1963). The Feynman lec- tures on physics (Reading Mass.: Addison-Wesley Pub. Co). Fitzgerald, D., Matusall, S., Skewes, J., and Roepstorff, A. (2014). What’s so critical about Critical Neuroscience? Rethinking experiment, enacting critique. Front. Hum. Neurosci. 8, 365. Friedman, W.A., Zeigler, H.P., and Keller, A. (2012). Vibrissae motor cortex unit activity during whisking. J. Neurophysiol. 107, 551–563. Fritsch, G., and Hitzig, E. (1870). Ueber die elektrische Erregbarkeit des Grosshirns. Arch. Anat. Physiol. Wiss. Med. 37, 300–332. Fuchs, E.C., Neitz, A., Pinna, R., Melzer, S., Caputi, A., and Monyer, H. (2016). Local and Distant Input Controlling Excitation in Layer II of the Medial Entorhinal Cortex. Neuron 89, 194–208. Gerdjikov, T. V, Haiss, F., Rodriguez-Sierra, O.E., and Schwarz, C. (2013). Rhythmic whisking area 17 Ebbesen (2017) Chapter 2

(RW) in rat primary motor cortex: an internal monitor of movement-related signals? J. Neuro- sci. 33, 14193–14204. Gioanni, Y., and Lamarche, M. (1985). A reappraisal of rat motor cortex organization by intracortical microstimulation. Brain Res. 344, 49–61. Giocomo, L.M., Moser, M.-B., and Moser, E.I. (2011). Computational models of grid cells. Neuron 71, 589–603. Grant, R. a., Sperber, A.L., and Prescott, T.J. (2012). The role of orienting in vibrissal touch sensing. Front. Behav. Neurosci. 6, 1–12. Graziano, M.S. a (2011). New insights into motor cortex. Neuron 71, 387–388. van Groen, T., and Wyss, J.M. (1990). The connections of presubiculum and parasubiculum in the rat. Hafting, T., Fyhn, M., Molden, S., Moser, M.-B., and Moser, E.I. (2005). Microstructure of a spatial map in the entorhinal cortex. Nature 436, 801–806. Hales, J.B., Schlesiger, M.I., Leutgeb, J.K., Squire, L.R., Leutgeb, S., and Clark, R.E. (2014). Medial entorhinal cortex lesions only partially disrupt hippocampal place cells and hippocampus-dependent place memory. Cell Rep. 9, 893–901. Hall, R.D., and Lindholm, E.P. (1974). Organization of motor and somatosensory neocortex in the albino rat. Brain Res. 66, 23–38. Hardy, L. (2001). Quantum Theory From Five Reasonable Axioms. arXiv 34. Harris, K.D. (2005). Opinion: Neural signatures of cell assembly organization. Nat. Rev. Neurosci. 6, 399–407. Harris, J.A., Hirokawa, K.E., Sorensen, S.A., Gu, H., Mills, M., Ng, L.L., Bohn, P., Mortrud, M., Ouellette, B., Kidney, J., et al. (2014). Anatomical characterization of Cre driver mice for neural circuit mapping and manipulation. Front. Neural Circuits 8, 76. Harris, K.D., Quiroga, R.Q., Freeman, J., and Smith, S.L. (2016). Improving data quality in neuronal population recordings. Nat. Neurosci. 19, 1165–1174. Hartley, T., Lever, C., Burgess, N., and O’Keefe, J. (2014). Space in the brain: how the hippocampal formation supports spatial cognition. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 369, 20120510. Hasselmo, M.E. (2005). What is the function of hippocampal theta rhythm?—Linking behavioral data to phasic properties of field potential and unit recording data. Hippocampus 15, 936–949. Heeger, D. (2006). Perception Lecture Notes: The Brain. [available: http://www.cns.nyu.edu/~david/ courses/perception/lecturenotes/brain/brain.html] Heldt, S.A., and Ressler, K.J. (2009). The Use of Lentiviral Vectors and Cre/loxP to Investigate the Function of Genes in Complex Behaviors. Front. Mol. Neurosci. 2, 22. 18 Ebbesen (2017) Introduction & Outline

Hill, D.N., Curtis, J.C., Moore, J.D., and Kleinfeld, D. (2011). Primary motor cortex reports efferent control of vibrissa motion on multiple timescales. Neuron 72, 344–356. Howard, M.W., and Eichenbaum, H. (2015). Time and space in the hippocampus. Brain Res. 1621, 345–354. Hubel, D.H., and Wiesel, T.N. (1959). Receptive fields of single neurones in the cat’s striate cortex. J. Physiol. 148, 574–591. Hubel, D.H., and Wiesel, T.N. (1998). Early exploration of the visual cortex. Neuron 20, 401–412. Kaufman, M.T., and Churchland, A.K. (2016). Many paths from state to state. Nat. Neurosci. 19, 1541–1542. Kim, C.K., Adhikari, A., and Deisseroth, K. (2017). Integration of optogenetics with complementary methodologies in systems neuroscience. Nat. Rev. Neurosci. 18, 222–235. Kitamura, T., Pignatelli, M., Suh, J., Kohara, K., Yoshiki, A., Abe, K., and Tonegawa, S. (2014). Island cells control temporal association memory. Science 343, 896–901. Kleinfeld, D., Berg, R.W., and O’Connor, S.M. (1999). Anatomical loops and their electrical dynamics in relation to whisking by rat. Somatosens. Mot. Res. 16, 69–88. Krakauer, J.W., Ghazanfar, A.A., Gomez-Marin, A., Maciver, M.A., and Poeppel, D. (2017). Neuron Perspective Neuroscience Needs Behavior: Correcting a Reductionist Bias. Neuron 93, 480–490. Krupic, J., Burgess, N., and O’Keefe, J. (2012). Neural Representations of Location Composed of Spatially Periodic Bands. Science (80-. ). 337, 853–857. Lehman, A.F. (1996). Measures of quality of life among persons with severe and persistent mental disorders. Soc. Psychiatry Psychiatr. Epidemiol. 31, 78–88. Lemon, R.N. (2008). An enduring map of the motor cortex. Exp. Physiol. 93, 798–802. McNaughton, B.L., Battaglia, F.P., Jensen, O., Moser, E.I., and Moser, M.-B. (2006). Path integration and the neural basis of the “cognitive map.” Nat. Rev. Neurosci. 7, 663–678. Molnár, Z., Kaas, J.H., de Carlos, J.A., Hevner, R.F., Lein, E., and Němec, P. (2014). Evolution and development of the mammalian cerebral cortex. Brain. Behav. Evol. 83, 126–139. Moser, E.I., Kropff, E., and Moser, M.-B. (2008). Place Cells, Grid Cells, and the Brain’s Spatial Rep- resentation System. Annu. Rev. Neurosci. 31, 69–89. Muller, U., Kubie, J.L., and Ranck, B. (1987). Spatial Firing Patterns Fixed Environment of Hip- pocampal Complex-Spike Cells in a. J. Neurosci. 7, 1935–1950. Neafsey, E.J., Bold, E.L., Haas, G., Hurley-Gius, K.M., Quirk, G., Sievert, C.F., and Terreberry, R.R. (1986). The organization of the rat motor cortex: A microstimulation mapping study. Brain Res. Rev. 11, 77–96. 19 Ebbesen (2017) Chapter 2

O’Connor, D.H., Huber, D., and Svoboda, K. (2009). Reverse engineering the mouse brain. Nature 461, 923–929. O’Keefe, J., and Dostrovsky, J. (1971). The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res. 34, 171–175. Penfield, W., and Rasmussen, T. (1952). The Cerebral Cortex of Man (New York: The Macmillan Company). Poucet, B., and Sargolini, F. (2013). A Trace of Your Place. Science (80) 340, 35–36. Ray, S., Naumann, R., Burgalossi, A., Tang, Q., Schmidt, H., and Brecht, M. (2014). Grid-layout and theta-modulation of layer 2 pyramidal neurons in medial entorhinal cortex. Science 343, 891–896. Rowland, D.C., Roudi, Y., Moser, M.-B., and Moser, E.I. (2016). Ten Years of Grid Cells. Annu. Rev. Neurosci. 39, 19–40. Sachdev, R.N.S., Sato, T., and Ebner, F.F. (2002). Divergent movement of adjacent whiskers. J. Neu- rophysiol. 87, 1440–1448. Sasaki, T., Leutgeb, S., and Leutgeb, J.K. (2014). Spatial and memory circuits in the medial entorhinal cortex. Curr. Opin. Neurobiol. 32C, 16–23. Shenoy, K. V, Sahani, M., and Churchland, M.M. (2013). Cortical Control of Arm Movements: A Dynamical Systems Perspective. Annu. Rev. Neurosci. 36, 337–359. Solstad, T., Boccara, C.N., Kropff, E., Moser, M.-B., and Moser, E.I. (2008). Representation of Geo- metric Borders in the Entorhinal Cortex. Science (80-. ). 322, 1865–1868. Taube, J.S., Muller, R.U., and Ranck, J.B. (1990). Head-direction cells recorded from the postsubiculum in freely moving rats. I. Description and quantitative analysis. J. Neurosci. 10, 420–435. Varga, C., Lee, S.Y., and Soltesz, I. (2010). Target-selective GABAergic control of entorhinal cortex output. Nat. Neurosci. 13, 822–824. Welker, W.I. (1964). Analysis of Sniffing of the Albino Rat. Behaviour 22, 223–244. Whissell, P.D., Tohyama, S., and Martin, L.J. (2016). The Use of DREADDs to Deconstruct Behav- ior. Front. Genet. 7, 70. WHO (1995). The World Health Organization quality of life assessment (WHOQOL): Position paper from the World Health Organization. Soc. Sci. Med. 41, 1403–1409. Witten, I.B., Steinberg, E.E., Lee, S.Y., Davidson, T.J., Zalocusky, K. a., Brodsky, M., Yizhar, O., Cho, S.L., Gong, S., Ramakrishnan, C., et al. (2011). Recombinase-driver rat lines: Tools, techniques, and optogenetic application to dopamine-mediated reinforcement. Neuron 72, 721–733. Wolfe, J., Mende, C., and Brecht, M. (2011). Social facial touch in rats. Behav. Neurosci. 125, 900– 20 Ebbesen (2017) Introduction & Outline

910. Wurtz, R.H. (2009). Recounting the impact of Hubel and Wiesel. J. Physiol. 587, 2817–2823. Yang, W., and Yuste, R. (2017). In vivo imaging of neural activity. Nat. Methods 14, 349–359. Zilli, E. a. (2012). Models of Grid Cell Spatial Firing Published 2005–2011. Front. Neural Circuits 6, 1–17. Žižek, S. (1989). The Sublime Object of Ideology (London: Verso). Zola-Morgan, S. (1995). Localization of Brain Function: The Legacy of Franz Joseph Gall (1758- 1828). Annu. Rev. Neurosci. 18, 359–383. Aaronson, S. (2004). Is Quantum Mechanics an Island in Theory Space. arXiv:quant-ph/0401062 1–9.

21 Ebbesen (2017) Chapter 4 4. Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex

This manuscript was published as: (*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Spicher, D. & Brecht, M. (2014) Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex. Neuron, 84(6):1191-1197. Reprinted with permission from Elsevier under CC BY-NC-ND Temporal discharge patterns of L2 MEC cells suggest that Grid cells ( ) are calbindin+ cells ( ) and Border cells ( ) are calbindin- cells ( ) Structure-function revealed by theta locking

Grid cells ≈ pyramidal cells

se a h p

a Locking t strength e h

0 π

Border cells ≈ stellate cells

Figure 8: Graphical abstract outlining the main message of this chapter: The temporal firing pattern of grid cells in superficial medial entorhinal cortex (blue dots, left) corresponds to the temporal firing pattern of pyramidal neurons (green dots, right). Similarly, the temporal firing pattern of border cells in superficial medial entorhinal cortex (red dots, left) match the temporal firing pattern of stellate neurons (black dots, right).

22 Ebbesen (2017) Layer 2 Grid & Border Cells

Neuron Report

Pyramidal and Stellate Cell Speciﬁcity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex

Qiusong Tang,1,3 Andrea Burgalossi,1,3,4,5,* Christian Laut Ebbesen,1,2,3 Saikat Ray,1 Robert Naumann,1,6 Helene Schmidt,1 Dominik Spicher,1 and Michael Brecht1,4,* 1Bernstein Center for Computational Neuroscience, Humboldt University of Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany 2Berlin School of Mind and Brain, Humboldt University of Berlin, Unter den Linden 6, 10099 Berlin, Germany 3Co-ﬁrst author 4Co-senior author 5Present address: Werner Reichardt Centre for Integrative Neuroscience, Otfried-Mu¨ ller-str. 25, 72076 Tu¨ bingen, Germany 6Present address: Max Planck Institute for Brain Research, Max-von-Laue-Str. 4, 60438 Frankfurt am Main, Germany *Correspondence: [email protected] (A.B.), [email protected] (M.B.) http://dx.doi.org/10.1016/j.neuron.2014.11.009 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).

SUMMARY the brain’s coordinate system supporting spatial navigation (see Moser and Moser, 2013 for review). In medial entorhinal cortex, layer 2 principal cells Pure grid cells are primarily found in layer 2 (Boccara et al., divide into pyramidal neurons (mostly calbindin 2010), which differs from other cortical laminae in its unique positive) and dentate gyrus-projecting stellate cells cell biology. Here the two types of principal cells, stellate and py- (mostly calbindin negative). We juxtacellularly labeled ramidal neurons, have been described (Alonso and Klink, 1993; layer 2 neurons in freely moving animals, but small Germroth et al., 1989). Specifically, stellate and pyramidal neu- sample size prevented establishing unequivocal rons differ in conductances and projection patterns (Alonso and Llina´ s, 1989; Lingenho¨ hl and Finch, 1991; Klink and Alonso, structure-function relationships. We show, however, 1997; Canto and Witter, 2012). Recent work indicates that stellate that spike locking to theta oscillations allows as- and pyramidal neurons can be reliably differentiated by calbindin signing unidentified extracellular recordings to pyra- immunoreactivity (Ray et al., 2014; Kitamura et al., 2014), and midal and stellate cells with 83% and 89% speci- that these cells also differ in their inhibitory inputs (Varga et al., ficity, respectively. In pooled anatomically identified 2010). Calbindin-positive (calbindin+) cells, which are clustered and theta-locking-assigned recordings, nonspatial and arranged in a hexagonal grid (Ray et al., 2014), have been discharges dominated, and weakly hexagonal spatial recently shown to project to the CA1 (Kitamura et al., 2014), while discharges and head-direction selectivity were calbindin-negative (calbindinÀ) neurons are homogeneously observed in both cell types. Clear grid discharges distributed and project primarily to the dentate gyrus (Varga were rare and mostly classified as pyramids (19%, et al., 2010; Ray et al., 2014). Few studies have so far explored 19/99 putative pyramids versus 3%, 3/94 putative structure-function relationships in entorhinal circuits (Schmidt- Hieber and Ha¨ usser, 2013; Domnisoru et al., 2013; Zhang et al., stellates). Most border cells were classified as stellate 2013; see Rowland and Moser, 2014 and Burgalossi and Brecht, (11%, 10/94 putative stellates versus 1%, 1/99 puta- 2014 for reviews). Thus, the functional implications of the remark- tive pyramids). Our data suggest weakly theta-locked able cellular diversity of layer 2 have remained largely unresolved. stellate border cells provide spatial input to dentate Resolving how differential spatial firing relates to principal gyrus, whereas strongly theta-locked grid discharges cell types will clarify the cellular mechanisms of grid discharges occur mainly in hexagonally arranged pyramidal cell and spatial input patterns to distinct subfields of the hippocam- patches and do not feed into dentate gyrus. pus. In the present work we aim at resolving layer 2 circuits by taking advantage of improved methodologies for identifying individual neurons recorded in freely moving animals. By cell identi- INTRODUCTION fication and theta-locking-based classification of unidentified recordings, we provide evidence that grid and border responses The medial entorhinal cortex is critically involved in spatial navi- are preferentially contributed by pyramidal and stellate cells, gation and memory. Among other functionally specialized cell respectively. types (Sargolini et al., 2006; Solstad et al., 2008; Savelli et al., 2008), it contains grid cells (Hafting et al., 2005), spatially modu- RESULTS lated neurons which show periodic, hexagonally arranged spatial firing fields. Given the striking regularity and invariance To explore the cellular basis of grid cell activity in medial entorhi- of the grid representation, these cells are thought to be part of nal cortex, we juxtacellularly recorded and labeled neurons in

Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors 1191

23 Ebbesen (2017) Chapter 4

Neuron Layer 2 Grid and Border Cells

Figure 1. Grid-like Firing Properties in a D A B C 16 Calbindin-Positive Pyramidal Neuron and L M Border Responses in a Calbindin-Negative

V Stellate Neuron (A) Top, micrograph (tangential section) of a calbindin+ neuron recorded in a rat exploring a 2D environment (1 3 1 m). Green, calbindin; red, neu-

Spike Count robiotin. Bottom, soma in red, green channel, and overlay. Scale bars, 100 mm (top), 10 mm (bottom). (B) Micrograph of a tangential layer 2 section with 0 360 720 calbindin immunoreactivity (green) and super- Theta Phase (Degree) imposed reconstruction of the pyramidal neuron D (white). The cell was poorly stained, basal dendrites 3.5 Hz 1.07 were minor, and a prominent apical dendrite extended toward the center of a calbindin patch ventral from the neuron. Scale bar, 250 mm. (C) Theta-phase histogram of spikes for the neuron shown in (A). For convenience, two repeated cycles are shown. The black sinusoid is a schematic local ﬁeld potential theta wave for reference. (D) Spike- trajectory plot, rate map, and 2D spatial autocorrelation of the rate map revealing the hexagonal grid cell periodicity. Spike-trajectory plot: red dots indicate spike locations, gray lines indicate the rat trajectory. Rate map: red indicates maximal ﬁring rate, value noted above. Spatial autocorrelation: color scale 1 (blue) through 0 (green) to 1 (red). For À EF G this cell, the grid score is 1.07. (E) Left, micrograph 22 (tangential section) of a calbindinÀ neuron recorded in a rat exploring a 2D environment (70 3 70 cm). Green, calbindin; red, neurobiotin. Right, soma in red, green channel, and overlay. Scale bars, 100 mm (left), 10 mm (right). (F) Micrograph of the tangential D

Spike Count layer 2 section with calbindin immunoreactivity L M (green) and superimposed reconstruction of the stellate neuron (white). The cell was well stained, V 0 360 720 and the huge dendritic field encompassed several Theta Phase (Degree) calbindin patches. Scale bar, 250 mm. (G) Theta- H 19.6 Hz 0.90 phase histogram of spikes for the neuron shown in (A). For convenience, two repeated cycles are shown. The black sinusoid is a schematic local field potential theta wave for reference. (H) Spike-trajectory plot, rate map, and 2D spatial autocorrelation of the rate map revealing the elongated firing field. Spike-trajectory plot: red dots indicate spike locations, gray lines indicate the rat trajectory. Rate map: red indicates maximal firing rate, value noted above. Spatial autocorrelation: color scale 0.5 À (blue) through 0 (green) to 0.5 (red). For this cell, the border score is 0.90. D, dorsal; L, lateral; M, medial; V, ventral.

layer 2 (which contains the largest percentage of pure grid cells; correlation analysis of the firing pattern in the 2D environment re- Boccara et al., 2010) in awake rats trained to explore 2D environ- vealed a hexagonal periodicity of firing fields (grid score = 1.07; ments (Tang et al., 2014). Clear grid cell discharges were rare. Figure 1D), indicative of grid cell activity (Hafting et al., 2005). The clearest grid-like firing pattern in our sample of 31 identified Because of its relatively low firing rate ( 0.5 Hz) this cell was cells (17 of which met the criteria for spatial analysis; see Exper- not included in the grid cell sample (see Experimental Proce- imental Procedures) was observed in the calbindin+ cell shown in dures). Most other identified calbindin+ neurons had no clear Figure 1A. This neuron had pyramidal morphology, with simple spatial firing patterns. dendritic arborization and a single large apical dendrite targeting The clearest border discharge in our sample of identified cells + a calbindin patch (Figure 1B; see also Ray et al., 2014). During was observed in the calbindinÀ cell shown in Figure 1E. This exploratory behavior, calbindin+ neurons fired with strong theta cell was a stellate neuron, which did not have a single apical rhythmicity and phase locked near the trough of the local field dendrite, but instead extended multiple and widely diverging potential theta rhythm (Figure 1C; Ray et al., 2014). Spatial auto- ascending dendrites; this dendritic tree spanned a vast field,

1192 Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors

24 Ebbesen (2017) Layer 2 Grid & Border Cells

Neuron Layer 2 Grid and Border Cells

which encompassed multiple calbindin+ patches (Figure 1F; result in a local and overlapping sampling of inputs in neighboring + see also Ray et al., 2014). On average, spikes from calbindinÀ calbindin cells, whereas neighboring calbindinÀ stellate cells neurons were weakly modulated by the local theta rhythm (Fig- sample large and nonoverlapping input territories. + ure 1G). In 3 out of 11 calbindinÀ cells from recordings with suf- CalbindinÀ stellate and calbindin pyramidal cells differ ficient spatial coverage, we observed clear border firing patterns strongly in their temporal discharge properties (Figures 1C and as in Figure 1H. While we did not observe grid cells, nonspatial 1G; Ray et al., 2014). We therefore wondered if temporal firing patterns also dominated in calbindinÀ neurons. discharge properties could be used to classify layer 2 cells as While the small size of the data set of identified neurons pre- putative pyramidal or stellate neurons. We used a support vector vented us from establishing firm structure-function relationships, machine to classify neurons based on both the spike phase and four preliminary observations can be drawn: (i) grid cells are less strength of phase locking to local field potential theta oscillations, + abundant in layer 2 than previously assumed (Sargolini et al., which indeed clearly segregated calbindin and calbindinÀ cells 2006; Boccara et al., 2010; but see Mizuseki et al., 2009; Gupta with a large distance to the separating hyperplane (Figure 2A; see et al., 2012; Bjerknes et al., 2014), and there is no one-to-one Supplemental Information). To further improve the purity of as- relationship between spatial discharge characteristics and cell signed cells, we added a guard zone around the hyperplane type, (ii) calbindin+ neurons probably include grid cells, (iii) the separating the Gaussian kernels classifying calbindin+ (light absence of grid cells in the 22 identified calbindinÀ stellate neu- green background) and calbindinÀ (gray background) cells (omit- rons suggests that grid cells are rare in this cell population, and ting the guard zone and classifying all cells did not qualitatively (iv) calbindinÀ neurons include border cells. affect the results; data not shown). We tested our classifier by a Currently available evidence points to a correspondence be- bootstrapping approach (Figures S2A and S2B) and found that + + tween cytochemical (calbindin versus calbindinÀ) and mor- a large fraction of calbindin and calbindinÀ cells could be phological (pyramidal versus stellate) classification of principal correctly assigned (Figure S2C). More importantly, the specificity neurons in layer 2 (Varga et al., 2010; Kitamura et al., 2014). To of classification procedure—reflected in the purity of the resulting further explore these relationships, we determined the percent- cell samples—was excellent, i.e., 89% for putative calbindinÀ age of calbindin+ cells in layer 2 and compared these data with cells and 83% for putative calbindin+ cells (Figure S2D), and related measurements in the literature (Figure S1A available on- even higher values for combination of identified and putatively line). In agreement with previous studies (Peterson et al., 1996; assigned cells (Figure S2E). We further evaluated the robustness Kumar and Buckmaster, 2006; Varga et al., 2010), we found of the classifier by testing it on a larger data set of identified layer that layer 2 neurons consist of 34% calbindin+ and 53% 2 neurons (Ray et al., 2014) recorded under urethane/ketamine + calbindinÀ (and reelin ) principal cells, and 13% interneurons anesthesia (Klausberger et al., 2003). We consider this a chal- (Figure S1B). We note that while Ray et al. (2014) found about lenging test of the classifier, as theta phase and strength of 30% of calbindin+ cells, most of which were shown to have pyra- locking might differ between the awake and anesthetized state. midal morphology (see also Varga et al., 2010; Kitamura et al., Similarly to the awake situation, however, the large majority of 2014), Gatome et al. (2010) found a slightly lower fraction of puta- neurons recorded under anesthesia were also correctly classi- + + tive pyramidal cells. Calbindin and calbindinÀ cells showed large fied (92% of calbindin cells, 65% of calbindinÀ cells, p < quantitative differences in their morphology, but without a clear 0.001, bootstrap; Figure 2B, bottom), suggesting that our classi- bimodality in individual morphological parameters (Figures S1C fication criteria work robustly and can effectively generalize and S1D). Calbindin+ cells had significantly (on average 2.5- across very different recording conditions (Figure 2B). Encour- fold) smaller dendritic trees (Figure S1E). Dendritic trees also aged by these results, we classified the larger data set of our hith- differed in shape between cell types. Calbindin+ cells had a single erto unidentified layer 2 juxtacellular and tetrode recordings long (always apical) dendrite, which accounted on average for (classified + identified n = 193 cells). 63% of the total dendritic length (Figure S1E) and which was To assess the relationship between cell identity and spatial polarized toward the center of pyramidal cell patches as shown firing properties, we pooled the nonidentified recordings, as- + previously (Ray et al., 2014). Calbindin expression matched signed to putative calbindin and calbindinÀ cells, with the re- well, but not perfectly, with pyramidal cell morphology (Figures cordings from histologically identified neurons. The pooled + S1C and S1D). CalbindinÀ cells featured similar-length dendrites data sets included n = 99 calbindin and n = 94 calbindinÀ cells, with the longest dendrite contributing on average for 33% of the respectively. In our first assessment of spatial discharge pat- total dendritic length (Figure S1E). These results are in line with terns, we attempted to classify grid and border cells solely using + published data and indicate that calbindin and calbindinÀ cells scores (grid score > 0.3, border score > 0.5; Solstad et al., 2008). largely correspond to pyramidal and stellate neurons, respec- According to visual inspection of individual rate maps, however, tively. However, the lack of clear morphological bimodality in layer these criteria were not sufficiently stringent and returned a 2 (see also Canto and Witter, 2012) implies that the correspon- majority of weakly to nonmodulated neurons, i.e., possibly a + dence between pyramidal/calbindin and stellate/calbindinÀ majority of false-positive grid and border cells. To resolve this might not be perfect. Interestingly, the spine density in calbindin+ issue, we adopted the cell classification approach of Bjerknes cells decreased as a function of distance from the soma, whereas et al. (2014), in which spatial discharge properties were only the reverse was true for calbindinÀ cells (Figure S1F). These quantified in those cells that carried significant amounts of morphological differences, together with clustering of calbindin+ spatial information (as assessed by a spike-shuffling procedure, cells in patches and the polarization of their apical dendrites see Skaggs et al., 1993; Supplemental Experimental Proce- toward the center of calbindin+ patches (Ray et al., 2014), likely dures). This approach identified grid and border responses,

Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors 1193

25 Ebbesen (2017) Chapter 4

Neuron Layer 2 Grid and Border Cells

A BC

DE F

G H

(legend on next page)

1194 Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors

26 Ebbesen (2017) Layer 2 Grid & Border Cells

Neuron Layer 2 Grid and Border Cells

A BC Figure 3. Temporal Spiking Properties of Grid Cells and Border Cells (A) Polar plot of theta strength as a function of preferred theta phase angle (4) for grid cells (blue dots) and border cells (red dots). (B) Theta strength of recorded grid cells (blue dots) and border cells (red dots) is significantly different. Blue and red lines indicate medians (p = 0.0013, Mann-Whitney U test). (C) Preferred theta phase for grid cells (blue dots) and border cells (red dots). Blue and red lines indicate circular means (p = 0.0000088, parametric Watson-Williams multisample test). Grid cells show a significant tendency to fire near the trough (p = 0.000000027, Rayleigh’s test for nonuniformity). Border cells show a tendency to fire near the peak of theta rhythm, but the phase locking to theta peak did not reach significance in our data set (p = 0.21, Rayleigh’s test for nonuniformity).

which in a majority of cases were convincing according to visual criteria, compared to only 3% (3/94) in the calbindinÀ population inspection. Consistent with previous studies (Hafting et al., 2005; (p = 0.00046, Fisher’s exact test). A higher fraction of calbindinÀ Sargolini et al., 2006; Boccara et al., 2010; Burgalossi et al., cells passed the border cell criterion (11% calbindinÀ, 10/94 2011; Domnisoru et al., 2013), a fraction of layer 2 neurons cells; versus 1% calbindin+, 1/99 cells), and this difference was (33%; n = 63 cells) were significantly spatially modulated. statistically significant (p = 0.0042, Fisher’s exact test). These Weak hexagonal symmetry of spatial firing patterns was data confirm and extend the conclusion from our recordings of + observed in both the calbindin and calbindinÀ data set, in line identified cells and indicate that grid cells are preferentially re- with previous observations (Burgalossi et al., 2011; Domnisoru cruited from the calbindin+ population, while border responses et al., 2013; Schmidt-Hieber and Ha¨ usser, 2013). However, preferentially occur in calbindinÀ cells. grid scores in the calbindin+ population were significantly higher Unlike many studies based on tetrode recordings (Sargolini than those in the calbindinÀ population (p = 0.000046, Mann- et al., 2006; Boccara et al., 2010; but see Zhang et al., 2013), a Whitney U test; Figures 2D and 2E), consistent with observations substantial fraction of cells showed head-direction selectivity + from the identified data set (Figure 1). On the other hand, in line both in identified and theta-assigned calbindin and calbindinÀ with observations from the identified data set (Figure 1), cells (Figure S4). Head-direction selectivity was more common + calbindinÀ cells had significantly higher border scores than in calbindin (19%, 19 out of 99 cells) than in calbindinÀ cells calbindin+ cells (Figure 2G; p = 0.0012, Mann-Whitney U test). (12%, 11 out of 94 cells), but this difference was not significant Border discharges in calbindinÀ cells are shown in Figure 2F, (p = 0.17, Fisher’s exact test), and both classes contained pure which also includes an example where border firing was as well as conjunctive responses (Sargolini et al., 2006). confirmed by a border test (Solstad et al., 2008; Lever et al., The grid and border cells recorded here showed systematic 2009). Thus, according to the grid and border scores shown in differences in spike locking to local field potential theta oscilla- Figures 2D and 2G, putative pyramidal and stellate cells have tions (Figure 3A). Spikes from most grid cells were strongly significantly different, but overlapping, spatial properties. entrained by the theta rhythm, with strong phase locking Figure 2H gives an overview of the spatial response properties (Figure 3B) and a phase preference near the theta trough (Fig- + of our pooled calbindin and calbindinÀ data sets, respectively ure 3C; p = 0.000000027, Rayleigh’s test for nonuniformity). (see also Figure S3). The majority of both calbindin+ and The modulation of spiking activity of border cells by the theta calbindinÀ neurons showed no significant spatial selectivity. rhythm was significantly weaker than in grid cells (Figure 3B; Grid patterns were significantly more common in the calbindin+ p = 0.0013, Mann-Whitney U test) and showed on average population, where 19% (19/99) of the cells passed our grid cell only a weak, nonsignificant phase preference for the theta

Figure 2. Cell Classification and Grid and Border Responses in Pooled Identified and Theta-Assigned Cells (A) Top, classification training set: polar plot of theta strength (value indicated by the upper-right number) and preferred theta phase angle (4) for calbindin+ cells (green dots) and calbindinÀ cells (black dots) identified in freely moving rats. Background color fill shows classification boundary based on 4 and theta strength; + cells in the pale green area and gray area will be classified as calbindin and calbindinÀ cells, respectively. Bottom, fraction of cells in classification categories. (B) + Top, polar plot of theta strength and of preferred theta phase angle (4) for calbindin cells (green dots) and calbindinÀ cells (black dots) identified in anaesthetized rats, overlaid on classification boundary. Bottom, fraction of cells in classification categories. (C) Top, polar plot of theta strength and preferred theta phase angle + (4) for nonidentified cells. Putative calbindin cells (red dots) and putative calbindinÀ cells (yellow dots) are shown overlaid on the classification boundary, together with unclassified cells (white dots). Bottom, estimate of the purity of the theta-assigned cell categories. The sample of putative calbindinÀ cells are estimated to be 89% pure, and the sample of putative calbindin+ cells are estimated to be 83% pure (see Figure S2). (D) Comparison of grid scores between (identified and + putative) calbindin and calbindinÀ neurons; the dotted line indicates the threshold for grid cell; vertical lines indicate medians (p = 0.000046, Mann-Whitney U test). (E) Representative grid firing pattern observed in calbindin+ neurons (spike-trajectory plot, rate map, and spatial autocorrelation; maximum firing rate and grid score indicated above plots). (F) Border firing patterns in calbindinÀ neurons. Conventions as in (E). Arrows indicate insertion of additional border. (G) + Comparison of border scores between (identified and putative) calbindin and calbindinÀ neurons; the dotted line indicates the threshold for border cells; vertical + lines indicate medians (p = 0.0012; Mann-Whitney U test). (H) Distribution of spatial discharge types in calbindin (left) and calbindinÀ (right) neurons was found to be significantly different in numbers of grid cells and border cells (p = 0.00046 and 0.0042, respectively, Pearson’s chi-square test). Cells that passed both grid and border score criteria were assigned to either grid or border cells, as specified in the Experimental Procedures.

Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors 1195

27 Ebbesen (2017) Chapter 4

Neuron Layer 2 Grid and Border Cells

peak (Figure 3C; p = 0.21, Rayleigh’s test for nonuniformity), cialists. Notably, the local similarity of grid cell discharges is which differed significantly from the phase preference of grid high, as neighboring grid cells share the same grid orientation cells (Figures 3B and 3C; p = 0.0000088, parametric Watson-Wil- and scaling and are phase coupled even across distinct environ- liams multisample test). Thus, in layer 2 grid and border signals ments (Hafting et al., 2005; Fyhn et al., 2007). We speculate that mirrored the temporal differences between calbindin+ pyramidal calbindin+ pyramidal neuron clustering and apical dendrite and calbindinÀ stellate cells reported earlier (Ray et al., 2014). bundling in patches (Ray et al., 2014) might impose this local similarity of grid discharges. A surprising implication of our data DISCUSSION is that the spatial input to the dentate gyrus is provided mainly by stellate border cells, whereas pyramidal grid cells do not Relating functionally defined discharge patterns to principal cell feed into this pathway (Kitamura et al., 2014; Ray et al., 2014). diversity is an unresolved issue in cortical physiology. In layer 2 of Border responses arise in stellate neurons, with long and widely medial entorhinal cortex, most studies suggested that spatially diverging dendritic trees, i.e., such discharge patterns may result modulated responses are common, and that grid firing patterns from a relatively global sampling of incoming inputs in medial en- are contributed by both stellate and pyramidal neurons (Burga- torhinal cortex and help generate place cell activity (Bjerknes lossi et al., 2011; Schmidt-Hieber and Ha¨ usser, 2013; Domnisoru et al., 2014; Bush et al., 2014). Recognizing the functional dichot- et al., 2013; Zhang et al., 2013). In line with such evidence, we omy of pyramidal and stellate cells in layer 2 will help elucidate observed a consistent fraction of spatially modulated neurons how spatial discharge patterns arise in cortical microcircuits. in layer 2, and weakly hexagonal firing patterns in both stellate and pyramidal neurons. At the same time, however, most grid EXPERIMENTAL PROCEDURES patterns that met our grid score and spatial information criteria (see Supplemental Experimental Procedures) were classified All experimental procedures were performed according to the German guidelines on animal welfare under the supervision of local ethics committees. Jux- as putative calbindin+ pyramidal cells (see Figure S3A). Border tacellular recordings and tetrode recordings in freely moving animals were responses, on the other hand, were predominantly observed in obtained in male Wistar and Long-Evans rats (150–250 g), which were habitu- the calbindinÀ stellate population (Figure S3B). Our data indicate ated to the behavioral arena and trained for 3–7 days. Experimental proce- a strong interdependence between cell type and spatial dures were performed as previously described (Burgalossi et al., 2011; Herfst discharge pattern in layer 2, where a calbindin+ cell is about six et al., 2012) with the exception that methodological developments allowed us times more likely to be a grid cell and ten times less likely to be to identify neurons in drug-free animals (Tang et al., 2014; see also Supple- mental Experimental Procedures). Some of the data have been published a border cell than a calbindinÀ neuron. Our confidence in classi- + in a previous report (Ray et al., 2014). Recordings in anesthetized animals fication is based on the striking differences between calbindin were performed under urethane/ketamine/xylazine (Klausberger et al., 2003). and calbindinÀ cells in their temporal discharge properties (Ray Juxtacellularly labeled neurons were visualized with streptavidin conjugated et al., 2014), the assessment of classification quality by our boot- to Alexa 546 (1:1,000). A Hilbert transform was used for assigning instanta- strapping approach, and the robustness of classification across neous theta phase of each spike based on theta in the local field potential in widely differing recording conditions. It is important, however, to the spike-theta phase analysis. The spatial periodicity of recorded neurons note that our conclusions rest on the validity and accuracy of our was assessed by spatial autocorrelations. Grid scores were calculated as previously described (Barry et al., 2012) by taking a circular sample of the spatial classification procedure. autocorrelogram, centered on, but excluding the central peak. To determine A key finding from our work is that layer 2 principal cells can be the modulation of a cell firing along a border, we determined border scores classified with high accuracy by their distinct temporal discharge as previously described or performed border tests (Solstad et al., 2008; Lever properties. Such classification can be extended to a large num- et al., 2009). Head-direction tuning was measured as the eccentricity of the cir- ber of unidentified layer 2 recordings from other laboratories, cular distribution of firing rates. Classification based on strength of locking to provided that the required histology and local field potential theta phase (S) and preferred theta phase angle (4) was done by building a support vector machine, trained on the vectors (cos( ),S,sin( ),S) using a data have been collected. To this end we provide our classifica- 4 4 Gaussian radial basis function kernel. Classification of nonidentified cells tion training data set (Table S1) and a custom-written MATLAB + into putative calbindin and calbindinÀ cells was performed by applying a con- function (Supplemental Information, Note S1). Such post hoc servative classification threshold, where we did not classify cells close to the assignment of principal cell types to recordings—i.e., supplying separating hyperplane. Detailed experimental and analytical procedures are identity to formerly blind extracellular recordings—could be provided in the Supplemental Information. instrumental for understanding principal cell diversity and cortical microcircuitry. SUPPLEMENTAL INFORMATION Calbindin+ pyramidal cells might be predetermined for grid cell Supplemental Information includes four figures, one table, and Supplemental function as they receive cholinergic inputs, are strongly theta Experimental Procedures and can be found with this article online at http://dx. modulated, and are arranged in a hexagonal grid (Ray et al., doi.org/10.1016/j.neuron.2014.11.009. 2014). We suggested an ‘‘isomorphic mapping hypothesis,’’ according to which an anatomical grid of pyramidal cells (Ray et al., AUTHOR CONTRIBUTIONS 2014) generates grid cell activity (Brecht et al., 2014) and is an embodiment of the brain’s representation of space in hexagonal Q.T. and A.B. performed juxtacellular recordings. C.L.E. and Q.T. performed tetrode recordings. S.R., R.N., and H.S. performed and analyzed anatomical grids. Representing grid discharge by a ‘‘cortical grid’’ might offer experiments. C.L.E. and D.S. developed the classifier and C.L.E. and Q.T. similar advantages as isomorphic representations of body parts, analyzed electrophysiology data. A.B. and M.B. conceived of the project as barrel fields (Woolsey and Van der Loos, 1970), or nose stripes and supervised experiments. All authors contributed to the writing of the (Catania et al., 1993), in somatosensory cortices of tactile spe- manuscript.

1196 Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors

28 Ebbesen (2017) Layer 2 Grid & Border Cells

Neuron Layer 2 Grid and Border Cells

ACKNOWLEDGMENTS Herfst, L., Burgalossi, A., Haskic, K., Tukker, J.J., Schmidt, M., and Brecht, M. (2012). Friction-based stabilization of juxtacellular recordings in freely moving This work was supported by Humboldt-Universita¨ t zu Berlin, BCCN Berlin rats. J. Neurophysiol. 108, 697–707. (German Federal Ministry of Education and Research BMBF, Fo¨ rderkennzei- Kitamura, T., Pignatelli, M., Suh, J., Kohara, K., Yoshiki, A., Abe, K., and chen 01GQ1001A), NeuroCure, the Neuro-Behavior ERC grant, and the Gott- Tonegawa, S. (2014). Island cells control temporal association memory. fried Wilhelm Leibniz Prize of the DFG. We thank Moritz von Heimendahl for Science 343, 896–901. programming, and Andreea Neukirchner, Juliane Steger, John Tukker, and Klausberger, T., Magill, P.J., Ma´ rton, L.F., Roberts, J.D., Cobden, P.M., Undine Schneeweiß. We are thankful to Francesca Sargolini, May-Britt and Buzsa´ ki, G., and Somogyi, P. (2003). Brain-state- and cell-type-specific firing Edvard Moser, and the other authors (Sargolini et al., 2006), who generously of hippocampal interneurons in vivo. Nature 421, 844–848. provided access to grid cell data, which were helpful in assembling an earlier version of this manuscript. Klink, R., and Alonso, A. (1997). Muscarinic modulation of the oscillatory and repetitive firing properties of entorhinal cortex layer II neurons. Accepted: November 6, 2014 J. Neurophysiol. 77, 1813–1828. Published: December 4, 2014 Kumar, S.S., and Buckmaster, P.S. (2006). Hyperexcitability, interneurons, and loss of GABAergic synapses in entorhinal cortex in a model of temporal REFERENCES lobe epilepsy. J. Neurosci. 26, 4613–4623. Lever, C., Burton, S., Jeewajee, A., O’Keefe, J., and Burgess, N. (2009). Alonso, A., and Llina´ s, R.R. (1989). Subthreshold Na+-dependent theta-like Boundary vector cells in the subiculum of the hippocampal formation. rhythmicity in stellate cells of entorhinal cortex layer II. Nature 342, 175–177. J. Neurosci. 29, 9771–9777. Alonso, A., and Klink, R. (1993). Differential electroresponsiveness of stellate Lingenho¨ hl, K., and Finch, D.M. (1991). Morphological characterization of rat and pyramidal-like cells of medial entorhinal cortex layer II. J. Neurophysiol. entorhinal neurons in vivo: soma-dendritic structure and axonal domains. 70, 128–143. Exp. Brain Res. 84, 57–74. Barry, C., Ginzberg, L.L., O’Keefe, J., and Burgess, N. (2012). Grid cell firing Mizuseki, K., Sirota, A., Pastalkova, E., and Buzsa´ ki, G. (2009). Theta oscilla- patterns signal environmental novelty by expansion. Proc. Natl. Acad. Sci. tions provide temporal windows for local circuit computation in the entorhi- USA 109, 17687–17692. nal-hippocampal loop. Neuron 64, 267–280. Bjerknes, T.L., Moser, E.I., and Moser, M.-B. (2014). Representation of geo- Moser, E.I., and Moser, M.-B. (2013). Grid cells and neural coding in high-end metric borders in the developing rat. Neuron 82, 71–78. cortices. Neuron 80, 765–774. Boccara, C.N., Sargolini, F., Thoresen, V.H., Solstad, T., Witter, M.P., Moser, Peterson, D.A., Lucidi-Phillipi, C.A., Murphy, D.P., Ray, J., and Gage, F.H. E.I., and Moser, M.-B. (2010). Grid cells in pre- and parasubiculum. Nat. (1996). Fibroblast growth factor-2 protects entorhinal layer II glutamatergic Neurosci. 13, 987–994. neurons from axotomy-induced death. J. Neurosci. 16, 886–898. Brecht, M., Ray, S., Burgalossi, A., Tang, Q., Schmidt, H., and Naumann, R. (2014). An isomorphic mapping hypothesis of the grid representation. Philos. Ray, S., Naumann, R., Burgalossi, A., Tang, Q., Schmidt, H., and Brecht, M. Trans. R. Soc. Lond. B Biol. Sci. 369, 20120521. (2014). Grid-layout and theta-modulation of layer 2 pyramidal neurons in medial entorhinal cortex. Science 343, 891–896. Burgalossi, A., and Brecht, M. (2014). Cellular, columnar and modular organization of spatial representations in medial entorhinal cortex. Curr. Opin. Rowland, D.C., and Moser, M.-B. (2014). From cortical modules to memories. Neurobiol. 24, 47–54. Curr. Opin. Neurobiol. 24, 22–27. Burgalossi, A., Herfst, L., von Heimendahl, M., Fo¨ rste, H., Haskic, K., Schmidt, Sargolini, F., Fyhn, M., Hafting, T., McNaughton, B.L., Witter, M.P., Moser, M., and Brecht, M. (2011). Microcircuits of functionally identified neurons in the M.-B., and Moser, E.I. (2006). Conjunctive representation of position, direction, rat medial entorhinal cortex. Neuron 70, 773–786. and velocity in entorhinal cortex. Science 312, 758–762. Bush, D., Barry, C., and Burgess, N. (2014). What do grid cells contribute to Savelli, F., Yoganarasimha, D., and Knierim, J.J. (2008). Influence of boundary place cell firing? Trends Neurosci. 37, 136–145. removal on the spatial representations of the medial entorhinal cortex. Hippocampus 18, 1270–1282. Canto, C.B., and Witter, M.P. (2012). Cellular properties of principal neurons in the rat entorhinal cortex. II. The medial entorhinal cortex. Hippocampus 22, Schmidt-Hieber, C., and Ha¨ usser, M. (2013). Cellular mechanisms of spatial 1277–1299. navigation in the medial entorhinal cortex. Nat. Neurosci. 16, 325–331. Catania, K.C., Northcutt, R.G., Kaas, J.H., and Beck, P.D. (1993). Nose stars Skaggs, W.E., McNaughton, B.L., Gothard, K.M., and Markus, E.J. (1993). An and brain stripes. Nature 364, 493. information-theoretic approach to deciphering the hippocampal code. In Domnisoru, C., Kinkhabwala, A.A., and Tank, D.W. (2013). Membrane potential Advances in Neural Processing Systems, Volume 5, S.J. Hanson, J.D. dynamics of grid cells. Nature 495, 199–204. Cowan, and C.L. Giles, eds. (San Mateo: Morgan Kaufmann). Fyhn, M., Hafting, T., Treves, A., Moser, M.-B., and Moser, E.I. (2007). Solstad, T., Boccara, C.N., Kropff, E., Moser, M.-B., and Moser, E.I. (2008). Hippocampal remapping and grid realignment in entorhinal cortex. Nature Representation of geometric borders in the entorhinal cortex. Science 322, 446, 190–194. 1865–1868. Gatome, C.W., Slomianka, L., Lipp, H.P., and Amrein, I. (2010). Number esti- Tang, Q., Brecht, M., and Burgalossi, A. (2014). Juxtacellular recording and mates of neuronal phenotypes in layer II of the medial entorhinal cortex of morphological identification of single neurons in freely moving rats. Nat. rat and mouse. Neuroscience 170, 156–165. Protoc. 9, 2369–2381. Germroth, P., Schwerdtfeger, W.K., and Buhl, E.H. (1989). Morphology of Varga, C., Lee, S.Y., and Soltesz, I. (2010). Target-selective GABAergic control identified entorhinal neurons projecting to the hippocampus. A light micro- of entorhinal cortex output. Nat. Neurosci. 13, 822–824. scopical study combining retrograde tracing and intracellular injection. Woolsey, T.A., and Van der Loos, H. (1970). The structural organization of layer Neuroscience 30, 683–691. IV in the somatosensory region (SI) of mouse cerebral cortex. The description Gupta, K., Keller, L.A., and Hasselmo, M.E. (2012). Reduced spiking in entorhi- of a cortical field composed of discrete cytoarchitectonic units. Brain Res. 17, nal cortex during the delay period of a cued spatial response task. Learn. Mem. 205–242. 19, 219–230. Zhang, S.-J., Ye, J., Miao, C., Tsao, A., Cerniauskas, I., Ledergerber, D., Hafting, T., Fyhn, M., Molden, S., Moser, M.-B., and Moser, E.I. (2005). Moser, M.-B., and Moser, E.I. (2013). Optogenetic dissection of entorhinal-hip- Microstructure of a spatial map in the entorhinal cortex. Nature 436, 801–806. pocampal functional connectivity. Science 340, 1232627.

Neuron 84, 1191–1197, December 17, 2014 ª2014 The Authors 1197

29 Ebbesen (2017) Chapter 4

Neuron, Volume 84

Supplemental Information

Pyramidal and Stellate Cell Specificity

of Grid and Border Representations

in Layer 2 of Medial Entorhinal Cortex Qiusong Tang, Andrea Burgalossi, Christian Laut Ebbesen, Saikat Ray, Robert Naumann, Helene Schmidt, Dominik Spicher, and Michael Brecht

30 Ebbesen (2017) Layer 2 Grid & Border Cells

Inventory list of supplemental data

Title Related to

Figure S1 Anatomical characterization of calbindin-positive Figure 1 pyramidal and calbindin-negative stellate cells in layer 2 of medial entorhinal cortex

Figure S2 Testing of the classifier and error estimates Figure 2

Figure S3 Firing properties of those identified and theta-assigned Figure 2 & 3 calbindin-positive and calbindin-negative neurons, which carry significant spatial information

Figure S4 Head-direction tuning of identified and theta-assigned Figure 2 calbindin-positive and calbindin-negative neurons

Table S1 Theta modulation (phase, strength) of juxtacellularly Figure 2 identified calbindin+ and calbindin- cells from layer 2 of medial entorhinal cortex

Note S1 Custom-written Matlab script to classify non- Figure 2 identified cells into putative calbindin+ and calbindin- by strength and preferred phase of theta locking

31 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S1, related to Figure 1

32 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S1. Anatomical characterization calbindin-positive pyramidal and calbindin- negative stellate cells in layer 2 of medial entorhinal cortex. (A) Percentage of all neurons classified as calbindin+, reelin+ (calbindin-) and interneurons in layer 2 of the medial entorhinal cortex of a rat. Absolute reelin+ numbers have been extrapolated from calbindin+ and reelin+ counts. Circular markers indicate measurement made in this study. Square markers indicate measurements made by other studies (Peterson et al., 1996; Kumar and Buckmaster, 2006; Varga et al., 2010). (B) Distribution of calbindin+ neurons (green), reelin+ neurons (black) and interneurons (red) in layer 2 of the medial entorhinal cortex of a rat. Numbers indicate averages of measurements indicated in panel A. (C) Identified calbindin+ cells have pyramidal morphologies. All cells come from tangential sections and are hence shown in ‘views from the top’. (D) Identified calbindin- cells have stellate morphologies. All cells come from tangential sections and are hence shown in ‘views from the top’. (E) Identified and reconstructed calbindin+ cells and calbindin- cells show significant (t-test) size and shape differences. (F) Spine distribution differs in calbindin+ cells and calbindin- cells; data refer to ten cells each, for which we counted spine densities in multiple ~30 µm dendrite segments at the distances from the soma specified in the plot. Slopes of spine density differed significantly between calbindin+ cells and calbindin- cells (p = 0.0023, t-test). Error bars indicate SEM.

33 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S2, related to Figure 2

34 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S2. Testing of the classifier and error estimates. (A) Theta strength and preferred theta phase of identified calbindin+ cells (green dots) and calbindin- cells (black dots) is significantly different. Green and black lines indicate medians of theta strength (p = 0.0033, Mann-Whitney U-test). Green and black lines indicate circular means of preferred theta phase (p = 0.000028, Parametric Watson-Williams multi-sample test). Calbindin+ cells show a significant tendency to fire near the trough (p = 0.013, Rayleigh's test for nonuniformity) and calbindin- cells show a tendency to fire near the peak of theta rhythm (p = 0.033, Rayleigh's test for nonuniformity). (B) Distribution of testing data for estimation of classifier performance. Testing data is generated by fitting the appropriate probability density functions (beta distributions and circular Gaussian distributions, respectively) to the distributions of theta strength and preferred theta phase of identified calbindin+ and calbindin- cells (N = 10.000 for both cell types). (C) Result of classification of testing data shows that both calbindin+ cells and calbindin- cells are classified with high accuracy and low false classification rates (75.5% correct and 17.5% incorrect for calbindin+ cells, 86.7% correct and 9.8% incorrect for calbindin- cells). This shows that the classification boundary is robust and not just overfitting the small training set of identified cells. (D) Estimation of the purity (positive predictive value) of the classifier based on the estimate of 34% calbindin+, 53% reelin+ (calbindin-) and 13% interneurons in L2 of rat MEC (Figure S1A,B). The sample of 93 putative calbindin+ cells is estimated to be 83% pure, and the sample of 83 putative calbindin- cells is estimated to be 89% pure. (E) After addition of identified, full-coverage cells (11 calbindin- and 6 calbindin+), we estimate the purity of our final cell sample to be 84% for calbindin+ cell and 90% for calbindin- cells.

35 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S3, related to Figure 2 & 3 A

36 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

37 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S3. Firing properties of those identified and theta-assigned calbindin-positive and calbindin-negative neurons, which carry significant spatial information. (A) Calbindin-positive neurons. Cells are ordered according to spatial firing properties. From left to right we show spike-trajectory plot, rate map, two-dimensional spatial autocorrelation and angular tuning (which is shown only for head-direction selective cells. Note that pure head- direction cells that do not carry positional information are not included in this Figure). Numbers above the rate map indicate maximum firing rate. Numbers above the spatial autocorrelation indicate grid or border scores with respect to their properties. (T) indicates cells recorded with tetrodes; all other cells are from juxtacellular recordings. (1) indicates cells recorded in 0.7 x 0.7 m arena, all other recordings are from 1 x 1 m arena. (B) Calbindin-negative neurons. Cells are ordered according to spatial firing properties. From left to right we show spike-trajectory plot, rate map, two-dimensional spatial autocorrelation and angular tuning (which is shown only for head-direction selective cells. Note that pure head- direction cells that do not carry positional information are not included in this Figure). Numbers above the rate map indicate maximum firing rate. Numbers above the spatial autocorrelation indicate grid or border scores with respect to their properties. (T) indicates cells recorded with tetrodes; all other cells are from juxtacellular recordings. (1) indicates cells recorded in 0.7 x 0.7 m arena, (2) one cell recorded in a 0.6 x 0.8 m arena; all other recordings are from 1 x 1 m arena.

38 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S4, related to Figure 2

39 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Figure S4. Head-direction tuning of those identified and theta-assigned calbindin-positive and calbindin-negative neurons. (A) Polar plots of the head-direction tuning in identified and theta-assigned calbindin-positive neurons, which carry significant directional information. Cells are ranked according to Rayleigh vector lengths. (T) indicates cells recorded with tetrodes; all other cells are from juxtacellular recordings. (1) indicates cells recorded in 0.7 x 0.7 m arena, all other recordings are from 1 x 1 m arena. (B) Polar plots of the head-direction tuning in identified and theta-assigned calbindin-negative neurons, which carry significant directional information. Cells are ranked according to Rayleigh vector lengths. (C) Comparison of HD index (Rayleigh vector length) between (identified and putative) calbindin+ and calbindin- neurons; the dotted line indicates the threshold for head-direction cell; vertical lines indicate medians (p = 0.835, Mann-Whitney U-test). (D) Numbers of head-direction cells in (identified and putative) calbindin+ (A) and calbindin- (B) neurons were not different (p = 0.17, Fisher’s exact test).

40 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Table S1, related to Figure 2

Table S1. Classification training dataset of putative calbindin+ cells or calbindin- cells. Cells were recorded and identified juxtacellularly in freely-moving animals. Phase value is the preferred firing phase in radians, in relation to theta rhythm. Strength is the average Rayleigh vector length of the phase locking to theta (0 to 1). Type = 1 means calbindin+, Type = 0 means calbindin-.

41 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Note S1, related to Figure 2

42 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

43 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

44 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Note S1. MatLab code of the function to classify unidentified cells as putative calbindin+ cells or calbindin- cells. Output of the function is CellClass and Distance. CellClass = 1 means calbindin+, CellClass = 0 means calbindin-. Distance is the signed distance to the classification boundary.

45 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Supplemental Experimental Procedures

All experimental procedures were performed according to German guidelines on animal welfare.

Freely-moving juxtacellular recordings Experimental procedures for obtaining juxtacellular recordings in freely moving animals were performed similar to earlier publications (Ray et al., 2014; Tang et al., 2014). Briefly, recordings were made from male Wistar and Long-Evans rats (150-350 g) maintained in a 12-h light / dark phase and were recorded in the dark phase. Glass pipettes with resistance 4-6 MΩ were filled with extracellular (Ringer) solution containing (in mM) NaCl 135, KCl 5.4, HEPES 5, CaCl2 1.8, and MgCl2 1 (pH = 7.2) and Neurobiotin (1-2%). Animal implantations were performed as previously described (Burgalossi et al., 2011; Herfst et al., 2012, Tang et al., 2014), with a basic head-implant including a metal post for head-fixation and placement of a miniaturized preamplifier, a plastic ring and a protection cap (Herfst et al., 2012). In order to target the dorsalmost region of medial entorhinal cortex, a plastic ring was glued on the skull surface 0.2- 0.8 mm anterior to the transverse sinus and 4.5-5 mm lateral to the midline. After implantation, rats were allowed to recover from the surgery and were habituated to head-fixation for 3-5 days, as previously described (Houweling et al., 2008, Tang et al., 2014). Rats were trained in the experimental arena for 3-7 days (multiple sessions per day, 15-20 min duration each). Within the recording arena (70 x 70 cm or 1 x 1 m square black box, with a white cue card on the wall; 1 cell was recorded in a square arena, 60 cm x 80 cm), rats were trained to chase for chocolate or sugar pellets. Training was performed both before and after implantation (see below), or after implantation only. On the day of recording, under isoflurane anesthesia (1-3%), implants were completed, and an additional metal post was cemented, which served to anchor the miniaturized micromanipulator (Kleindiek Nanotechnik GmbH; Lee et al., 2006; Tang et al., 2014). 3-4 hours to 1 day later after recovery from anesthesia, rats were head-fixed, full implants were assembled, and the miniaturized micromanipulator and preamplifier were secured to the metal posts. The glass recording pipette was advanced into the brain; a thick agarose solution (4-5% in Ringer) was applied into the recording chamber for sealing the craniotomy and stabilization. Animals were then released and gently transferred into the behavioral arena. To minimize discomfort from the head implant, we sometimes supplied local anesthesia in the neck region. Juxtacellular recordings were established while animals were running in the arena. Juxtacellular labeling was attempted at the end of the recording session according to standard procedures (Pinault et al., 1996). A number of recordings (non-identified recordings; see data analysis) putatively in layer 2 (n = 61) were either lost before the labeling could be attempted, or the recorded neurons could not be unequivocally identified. After the experiment, the animals were euthanized with an overdose of ketamine or urethane and perfused transcardially with 0.1 M PB followed by 4% paraformaldehyde solution, shortly after the labeling protocol. Juxtacellular recordings in anesthetized animals (Ray et al., 2014) were performed under ketamine/urethane anesthesia according to established procedures (Klausberger et al., 2003, Quilichini et al., 2010). The juxtacellular signals were amplified by the ELC-03XS amplifier and sampled at 20 kHz by a data-acquisition interface under the control of PatchMaster 2.20 software. The animal’s location and head-direction was automatically tracked at 25 Hz by the Neuralynx video tracking system and two head-mounted LEDs.

Tetrode recordings Tetrode recordings (n = 126 layer 2 single units) were obtained as previously described in detail (von Heimendahl et al., 2012). Tetrodes were turned from 12.5 μm diameter nichrome wire

46 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

(California Fine Wire Company) and goldplated to ~250 kOhm impedance. Spiking activity and local field potential were recorded at 32 kHz (Neuralynx; Digital Lynx). Local field potential for theta phase assignment was recoded from the same tetrode as single units, relative to one tetrode left in superficial cortex. All recordings were done in a 1x1m box with behavioral training tasks same as juxtacellular procedures. The animal’s location and head-direction was automatically tracked at 25 Hz by video tracking and head-mounted LEDs, as described above. After recordings, tetrode tracks were lesioned and the animal was trancardially perfused. The brain was sectioned tangentially and recording sites assigned by histology. Spikes were pre-clustered using KlustaKwik (K.D. Harris, Rutgers University) and manually using MClust (A.D. Redish, University of Minnesota). Cluster quality was assessed by spike shape, ISI-histogram, L-ratio and isolation distance, as previously described (von Heimendahl et al., 2012). Putative interneurons were identified based on firing rate, spike shape and ISI-histogram and were excluded from classification.

Neurobiotin labeling and calbindin immunohistochemistry For histological analysis of juxtacellularly-labeled neurons, Neurobiotin was visualized with streptavidin conjugated to Alexa 546 (1:1000). Subsequently, immunohistochemistry for Calbindin was performed as previously described (Ray et al., 2014) and visualized with Alexa Fluor 488. After fluorescence images were acquired, the Neurobiotin staining was converted into a dark DAB reaction product. Neuronal morphologies were reconstructed by computer-assisted manual reconstructions (Neurolucida).

Spine density measurement To assess the spine density of calbindin+ and calbindin- dendrites, we labeled neurons in vivo juxtacellularly and identified the cells based on their calbindin immunoreactivity. We counted spines of fluorescent and DAB converted cells (10 calbindin+ and 10 calbindin- neurons) at 50 µm, 100 µm and 150 µm from the soma. The spine counts were normalized by dendritic length to obtain the number of spines per µm.

Estimate of the fraction of unintentionally included non-layer 2 cells Targeting recordings to layer 2 was achieved by mapping (1) the depth at which a pronounced increase in spiking activity and multi-unit synchrony during running was observed (Domnisoru et al., 2013) and (2) the L2/L1 border, which was always easy to identify as a reference point due to the drop in spiking activity and the more prominent local field potential gamma oscillations during theta epochs observed in L1 (Quilichini et al., 2010). We estimated the fraction of non-layer 2 principal cells included in our sample and expect that this mistaken fraction is in the < 10% range and probably consists mainly of parasubicular cells. This estimate was computed as follows: (1) We included 126 unidentified cells from tetrodes, and according to histology this sample does not contain off-target cells, as all recording sites could be reliably assigned to medial entorhinal cortex layer 2. (2) The laminar mistakes in our assignment of juxtacellular recordings seem to be small, i.e. in 31 recording attempts where we aimed at layer 2, we never recovered layer 6, layer 5 and layer 4 cells, but indeed recovered cells in the expected target. In only 2 additional cases we recovered cells in layer 3, where we expected to find layer 2 cells. Hence we expect a 6% laminar error rate. (3) Dorsoventral / mediolateral mistakes. We never recovered unintentionally postrhinal, retrospenial or lateral entorhinal cells in our medial entorhinal cortex recording attempts. However, in 9 experiments where we aimed at targeting layer 2, 9 parasubicular cells were

47 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

recovered instead. Thus, there are probably also parasubicular cells in the unidentified cells sample and this error might appear to be substantial (22% error rate). However, in 36 of the 61 included unidentified cells we could exclude such mistakes, because we identified the respective tracks in the correct target location. In 25 included unidentified juxtacellular recordings we could not rule out such mistakes, because tracks were not found, because of poor histology or proximity of tracks to the parasubiculum. From these numbers we expect about 4 laminar mistakes (unintentionally recorded layer 3 cells) in the 61 included juxtacellular recordings. We expect about 6 dorsoventral / mediolateral mistakes (unintentionally recorded parasubicular cells) in the 25 included juxtacellular recordings, where we could not exclude such mistakes. This leads to the following overall numbers: 16% unidentified recordings are expected to be non- layer 2 cells. This corresponds to a 5% rate of mistakes in our overall sample (identified and unidentified cells).

Analysis of theta locking For all cells, we calculated the locking to theta phase based on spiking discharge in relation to theta rhythm in the local field potential. The local field potential was zero-phase band-pass filtered (4-12 Hz) and a Hilbert transform was used to determine the instantaneous phase of the theta wave. The strength of locking to theta phase, S, and the preferred phase angle, φ, was defined as the modulus and argument of the Rayleigh average vector of the theta phase at all spike times. Only spikes during running (speed cutoff = 1 cm/s for juxtacellular signals, 5 cm/s for tetrode recordings) were included in the analysis. Only cells with firing rate ≥ 0.5 Hz were included in the analysis (Barry et al., 2012b). Both the analysis procedures and the juxtacellular data set largely correspond to our recent publication (Ray et al., 2014), whereby a more stringent band-pass filtering was applied in a subset of cells.

Analysis of Spatial Modulation The position of the rat was defined as the midpoint between two head-mounted LEDs. A running speed threshold (see above) was applied for isolating periods of rest from active movement. Color-coded firing maps were plotted. For these, space was discretized into pixels of 2.5 cm x 2.5 cm, for which the occupancy z of a given pixel x was calculated as

z(x)= ∑ w(|x − xt |)Δt t where xt is the position of the rat at time t, Δt the inter-frame interval, and w a Gaussian smoothing kernel with σ = 5cm. Then, the firing rate r was calculated as

∑ w(|x − xi |) r(x)= i z where xi is the position of the rat when spike i was fired. The firing rate of pixels, whose occupancy z was less than 20 ms, was considered unreliable and not shown. To determine the spatial periodicity of juxtacellularly recorded neurons, we determined spatial autocorrelations. The spatial autocorrelogram was based on Pearson’s product moment correlation coefficient:

48 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

where, ( , )the autocorrelation between pixels or bins with spatial offset τx and τy. f is the image without smoothing or the firing rate map after smoothing, n is the number of overlapping 𝑥𝑥 𝑦𝑦 pixels or𝑟𝑟 bins.𝜏𝜏 𝜏𝜏 Autocorrelations were not estimated for lags of τx and τy, where n < 20. For spatial and head-directional analysis, both a spatial (> 50% spatial coverage) and a firing rate inclusion criterion (> 0.5 Hz) were applied. Spatial coverage was defined as the fraction of visited pixels (bins) in the arena to the total pixels.

Analysis of Spatial Information For all cells, we calculated the spatial information rate, I, from the spike train and rat trajectory:

1 ( ) = ( ) log ( ) 𝑟𝑟 𝑥𝑥 𝐼𝐼 � 𝑟𝑟 𝑥𝑥 2 𝑜𝑜 𝑥𝑥 𝑑𝑑𝑥𝑥 where r(x) and o(x) are the firing rate𝑇𝑇 and occupancy𝑟𝑟 ̅as a function of a given pixel x in the rate map. is the overall mean firing rate of the cell and T is the total duration of a recording session (Skaggs et al., 1993). A cell was determined to have a significant amount of spatial information, if the 𝑟𝑟observed̅ spatial information rate exceeded the 95th percentile of a distribution of values of I obtained by circular shuffling. Shuffling was performed by a circular time-shift of the recorded spike train relative to the rat trajectory by a random time ]0, [ for 1000 permutations (von Heimendahl et al., 2012; Bjerknes et al., 2014). ′ 𝑡𝑡 ∈ 𝑇𝑇 Analysis of Gridness Grid scores were calculated as previously described (Barry et al., 2012a) by taking a circular sample of the autocorrelogram, centered on, but excluding the central peak. The Pearson correlation of this circle with its rotation for 60 degrees and 120 degrees was obtained (on peak rotations) and also for rotations of 30 degrees, 90 degrees and 150 degrees (off peak rotations). Gridness was defined as the minimum difference between the on-peak rotations and off-peak rotations. To determine the grid scores, gridness was evaluated for multiple circular samples surrounding the center of the autocorrelogram with circle radii increasing in unitary steps from a minimum of 10 pixels more than the width of the radius of the central peak to the shortest edge of the autocorrelogram. The radius of the central peak was defined as the distance from the central peak to its nearest local minima in the spatial autocorrelogram. The radius of the inner circle was increased in unitary steps from the radius of the central peak to 10 pixels less than the optimal outer radius. The grid score was defined as the best score from these successive samples. Grid scores reflect both the hexagonality in a spatial field and also the regularity of the hexagon. To disentangle the effect of regularity from this index, and consider only hexagonality, we transformed the elliptically distorted hexagon into a regular hexagon and computed the grid scores (Barry et al., 2012a). A linear affine transformation was applied to the elliptically distorted hexagon, to stretch it along its minor axis, until it lay on a circle, with the diameter equal to the major axis of the elliptical hexagon. The grid scores were computed on this transformed regular hexagon (Barry et al., 2012a).

Analysis of Border Cells

49 Ebbesen (2017) Chapter 4

Tang et al. Layer 2 Grid & Border Cells Supplemental information

To determine the modulation of a cell firing along a border, we determined border scores (Solstad et al., 2008). Border fields were identified from a collection of neighboring pixels having a firing rate higher than 0.3 times the maximum firing rate and covering an area of at least 100 cm (Sargolini et al., 2006). The coverage (Cm) along a wall was defined as the maximum length of a putative border field parallel to a boundary, divided by the length of the boundary. The mean firing distance (Dm) of a field was defined as the sum of the square of its distance from the boundary, weighted by the firing rate (Solstad et al., 2008). The distance from a boundary was defined as the exponential of the square of the distance in pixels from the closest boundary, normalized by half the length of the boundary. Border scores were defined as the maximum difference between Cm and Dm, divided by their sum, and ranged from -1 to +1.

Analysis of Head Direction Head-direction tuning was measured as the excentricity of the circular distribution of firing rates. For this, firing rate was binned as a function of head-direction (N = 36). A cell was said to have a significant head-direction tuning, if the length of the average vector exceeded the 95th percentile of a distribution of average vector lengths calculated from shuffled data and had a Rayleigh vector length > 0.3. Data was shuffled by applying a random circular time-shift to the recorded spike train for 1000 permutations.

Classification of non-identified cells into putative cell types For classification based on strength of locking to theta phase, S, and preferred theta phase angle, φ, we built a support vector machine using the built-in functions of the MATLAB Statistics Toolbox (The MathWorks Inc., Natick, MA, USA) using pairs of φ and S obtained from juxtacellular recording of identified cells. Because the phase angle is a circular variable, we trained the classifier on a space of the vectors (cos(φ)·S,sin(φ)·S), scaled to zero mean and unit variance using a gaussian radial basis kernel function with a scaling factor, sigma, of 1. To avoid cross-contamination of the two clusters, we employed a guard zone and excluded cells with a distance to the classification hyperplane < 0.1 from classification (white dots in Figure 2C). Classifier robustness was evaluated using a bootstrapping approach. To test if the putative calbindin+/calbindin- border suggested by the classifier based on our limited set of identified cells would also correctly classify a large number of non-identified cells, we fitted the appropriate probability density functions to the theta strength and phase angle of identified cells (beta distributions and circular Gaussian distributions, respectively) and generated 10.000 calbindin+ and 10.000 calbindin- testing cells drawn from these distributions (Figure S2A and S2B). Testing cells were classified and found to be classified 75.5% correctly for calbindin+ cells and 86.7 % correctly for calbindin- cells (Figure S2C), suggesting that our classifier generally performs well and is not just overfitting our small dataset of identified cells from freely-moving rats. Assuming the prior distribution of ~34% calbindin+ neurons , ~53% reelin+ (calbindin-) neurons and ~13% interneurons in layer 2 of the medial entorhinal cortex of a rat (Figure S1A and S1B), we estimate the purity (positive predictive value) of putative calbindin+ and putative calbindin- cells assigned by our classifier to be 83% and 89%, respectively (Figure S2D). This gives us the final cell sample purity of our putative and identified dataset of 84% and 90% for calbindin+ and calbindin- cells, respectively.

Classification of cells into functional categories Cells were classified as head-direction cells, grid cells, conjunctive cells, border cells and nonspatially modulated cells based on their grid score, border score, spatial information and significance of head-directionality according to the following criteria:

50 Ebbesen (2017) Layer 2 Grid & Border Cells

Tang et al. Layer 2 Grid & Border Cells Supplemental information

Head direction cells: Rayleigh vector length > 0.3 & significant head-direction tuning (Boccara et al., 2010) Grid cells: Grid score > 0.3 & significant spatial information. Border cells: Border score > 0.5 & significant spatial information (Solstad et al., 2008), or those who passed border test (Lever et al., 2009). Spatially irregular cells: significant spatial information (Bjerknes et al., 2014). Non-spatially modulated cell: no significant spatial information. In agreement with previous work (Solstad et al., 2008), few cells (n = 6) passed both the border cell and the grid cell threshold. These six cells were assigned to be grid cells by visual inspection.

Supplemental References

Barry, C., Bush, D., O'Keefe, J., and Burgess, N. (2012). Models of grid cells and theta oscillations. Nature 488, 103.

von Heimendahl, M., Rao, R.P., and Brecht, M. (2012). Weak and nondiscriminative responses to conspecifics in the rat hippocampus. J. Neurosci. 6, 2129-41.

Houweling, A.R., Doron, G., Voigt, B.C., Herfst, L.J., and Brecht, M. (2010). Nanostimulation: manipulation of single neuron activity by juxtacellular current injection. J. Neurophysiol. 103, 1696-704.

Lee, A.K., Manns, I.D., Sakmann, B., and Brecht, M. (2006). Whole-cell recordings in freely moving rats. Neuron 51, 399-407.

Pinault, D. (1996). A novel single-cell staining procedure performed in vivo under electrophysiological control: morpho-functional features of juxtacellularly labeled thalamic cells and other central neurons with biocytin or Neurobiotin. J. Neurosci. Methods 65, 113-136.

Quilichini, P., Sirota, A., and Buzsáki, G. (2010). Intrinsic circuit organization and theta-gamma oscillation dynamics in the entorhinal cortex of the rat. J. Neurosci. 18, 11128-42.

51 Ebbesen (2017) Chapter 5 5. Functional Architecture of the Rat Parasubiculum

This manuscript was published as: (*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., (*)Sanguinetti-Scheck, J.I., Schmidt, H., Tukker, J.J., Naumann, R., Ray, S., Preston-Ferrer, P., Schmitz, D., Brecht, M. (2016) Functional Architec- ture of the Rat Parasubiculum. Journal of Neuroscience 36(7):2289-2301. Reprinted with permission from Soc. for Neuroscience under CC Attribution 4.0 Intl. Licence

52 Ebbesen (2017) Functional Architecture of the Parasubiculum

The Journal of Neuroscience, February 17, 2016 • 36(7):2289–2301 • 2289

Systems/Circuits Functional Architecture of the Rat Parasubiculum

Qiusong Tang,1* Andrea Burgalossi,2* Christian Laut Ebbesen,1,3* Juan Ignacio Sanguinetti-Scheck,1* Helene Schmidt,1 X John J. Tukker,1,4 Robert Naumann,1 Saikat Ray,1 Patricia Preston-Ferrer,2 Dietmar Schmitz,4 and Michael Brecht1 1Bernstein Center for Computational Neuroscience, Humboldt Universita¨t zu Berlin, 10115 Berlin, Germany, 2Werner Reichardt Centre for Integrative Neuroscience, 72076 Tu¨bingen, Germany, 3Berlin School of Mind and Brain, Humboldt University of Berlin, 10115 Berlin, Germany, and 4Charite´ Universita¨tsmedizin Berlin, 10117 Berlin, Germany

The parasubiculum is a major input structure of layer 2 of medial entorhinal cortex, where most grid cells are found. Here we investigated parasubicular circuits of the rat by anatomical analysis combined with juxtacellular recording/labeling and tetrode recordings during spatial exploration. In tangential sections, the parasubiculum appears as a linear structure flanking the medial entorhinal cortex medi- odorsally. With a length of ϳ5.2 mm and a width of only ϳ0.3 mm (approximately one dendritic tree diameter), the parasubiculum is both one of the longest and narrowest cortical structures. Parasubicular neurons span the height of cortical layers 2 and 3, and we observed no obvious association of deep layers to this structure. The “superficial parasubiculum” (layers 2 and 1) divides into ϳ15 patches, whereas deeper parasubicular sections (layer 3) form a continuous band of neurons. Anterograde tracing experiments show that parasubicular neurons extend long “circumcurrent” axons establishing a “global” internal connectivity. The parasubiculum is a prime target of GABAergic and cholinergic medial septal inputs. Other input structures include the subiculum, presubiculum, and anterior thalamus. Functional analysis of identified and unidentified parasubicular neurons shows strong theta rhythmicity of spiking, a large fraction of head-direction selectivity (50%, 34 of 68), and spatial responses (grid, border and irregular spatial cells, 57%, 39 of 68). Parasubicular output preferentially targets patches of calbindin-positive pyramidal neurons in layer 2 of medial entorhinal cortex, which might be relevant for grid cell function. These findings suggest the parasubiculum might shape entorhinal theta rhythmicity and the (dorsoventral) integration of information across grid scales. Key words: anatomy; border cell; head-direction cell; medial entorhinal cortex; parasubiculum; spatial navigation

Significance Statement Gridcellsinmedialentorhinalcortex(MEC)arecrucialcomponentsofaninternalnavigationsystemofthemammalianbrain.The parasubiculum is a major input structure of layer 2 of MEC, where most grid cells are found. Here we provide a functional and anatomical characterization of the parasubiculum and show that parasubicular neurons display unique features (i.e., strong theta rhythmicity of firing, prominent head-direction selectivity, and output selectively targeted to layer 2 pyramidal cell patches of MEC). These features could contribute to shaping the temporal and spatial code of downstream grid cells in entorhinal cortex.

Introduction (Moser et al., 2008; Moser and Moser, 2013; Burgess, 2014). Ex- The analysis of spatial discharge patterns in hippocampal and tracellular recordings revealed an astonishing degree of complex- parahippocampal brain regions is a remarkable success story ity, abstractness, but also identified clear behavioral correlates of discharge patterns, such as place, head-direction, border, and grid cells. Along with the exploration of discharge properties, Received Oct. 13, 2015; revised Jan. 11, 2016; accepted Jan. 13, 2016. Author contributions: Q.T., A.B., C.L.E., J.I.S.-S., and M.B. designed research; Q.T., A.B., C.L.E., J.I.S.-S., H.S., J.J.T., anatomists delineated in great detail the basic circuitry of the R.N.,S.R.,P.P.-F.,D.S.,andM.B.performedresearch;Q.T.,A.B.,C.L.E.,J.I.S.-S.,H.S.,J.J.T.,R.N.,S.R.,P.P.-F.,D.S.,and M.B. analyzed data; Q.T., A.B., C.L.E., J.I.S.-S., H.S., J.J.T., R.N., S.R., P.P.-F., D.S., and M.B. wrote the paper. This work was supported by Humboldt-Universita¨t zu Berlin, BCCN Berlin (German Federal Ministry of Education andResearchBMBF,Fo¨rderkennzeichen01GQ1001A),NeuroCure,Neuro-BehaviorERCGrant,DeutscheForschungs- Correspondence should be addressed to either of the following: Dr. Andrea Burgalossi, Werner Reicha- gemeinschaft Gottfried Wilhelm Leibniz Prize, and the Werner Reichardt Centre for Integrative Neuroscience at the rdt Centre for Integrative Neuroscience, Otfried-Mu¨ller-Str. 25, 72076 Tu¨bingen, Germany, E-mail: EberhardKarlsUniversityofTu¨bingen.TheCentreforIntegrativeNeuroscienceisanExcellenceClusterfundedbythe [email protected]; or Dr. Michael Brecht, Bernstein Center for Computational Neuro- DeutscheForschungsgemeinschaftwithintheframeworkoftheExcellenceInitiative(EXC307).WethankMoritzvon science, Humboldt Universita¨t zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany, E-mail: Heimendahl for programming; Falko Fuhrmann and Stefan Remy for generous donations of PV-Cre mice and help [email protected]. with the AAV injections; Susanne Schoch (University of Bonn) for providing the virus; and Andreea Neukirchner, R. Naumann’s current address: Max-Planck-Institute for Brain Research, Max-von-Laue-Str. 4, 60438 Frankfurt Juliane Steger, Alexandra Eritja, Susanne Rieckmann, and Undine Schneeweiß for outstanding technical assistance. am Main, Germany. The authors declare no competing financial interests. DOI:10.1523/JNEUROSCI.3749-15.2016 *Q.T., A.B., C.L.E., and J.I.S.-S. contributed equally to this work. Copyright © 2016 the authors 0270-6474/16/362289-13$15.00/0

53 Ebbesen (2017) Chapter 5

2290 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

hippocampal formation (Amaral and Witter, 1989; van Strien et (nominal thickness 100 ␮m, horizontal) right after overnight fixation in al., 2009). PFA. The detailed data available about certain parts of the hip- Histochemistry and immunohistochemistry. Acetylcholinesterase activ- pocampal formation, such as dorsal CA1 in the rodent, should ity was visualized according to previously published procedures (Ray et not blind us for gaps in our knowledge about less “classic” hip- al., 2014). After washing brain sections in a solution containing 1 ml of 0.1 M citrate buffer, pH 6.2, and 9 ml 0.9% NaCl saline solution pocampal processing nodes. The parasubiculum is one such (CS), sections were incubated with CS containing 3 mM CuSO , 0.5 structure that lies beyond the classic trisynaptic hippocampal 4 mMK3Fe(CN)6, and 1.8 mM acetylthiocholine iodide for 30 min. After loop (Andersen et al., 1971) and has been investigated relatively rinsing in PB, reaction products were visualized by incubating the sec- little. This parahippocampal region provides massive input to tions in PB containing 0.05% 3,3Ј- DAB and 0.03% nickel ammonium layer 2 of medial entorhinal cortex (van Groen and Wyss, 1990; sulfate. Immunohistochemical stainings were performed according to Caballero-Bleda and Witter, 1993, 1994) and shows prominent standard procedures. Briefly, brain sections were preincubated in a expression of markers for cholinergic activity (Slomianka and blocking solution containing 0.1 M PBS, 2% BSA, and 0.5% Triton X-100 Geneser, 1991). Early physiological analysis described a small (PBS-X) for an hour at room temperature. Following this, primary anti- fraction of place-responsive cells in the parasubiculum (Taube, bodies were diluted in a solution containing PBS-X and 1% BSA. We 1995), and subsequent extracellular recordings have also identi- used primary antibodies against the calcium binding protein Calbindin fied head-direction, border, and grid responses among parasu- (1:5000), the DNA binding neuron-specific protein NeuN (1:1000), and, for the mice, against GFP. Incubations with primary antibodies were bicular neurons (Cacucci et al., 2004; Boccara et al., 2010). allowed to proceed for at least 24 h under mild shaking at 4°C in free- From both a physiological and an anatomical perspective, the floating sections. Incubations with primary antibodies were followed by parasubiculum is somewhat difficult to study. First, the small size detection with secondary antibodies coupled to different fluorophores of the parasubiculum complicates recordings and tracer injec- (Alexa-488 and Alexa-546). Secondary antibodies were diluted (1:500) in tions. Second, the parasubicular position (on the caudal edge of PBS-X, and the reaction was allowed to proceed for2hinthedark at the parahippocampal lobe wrapping around entorhinal cortex, room temperature. For multiple antibody labeling, antibodies raised in which goes along with a strong bending of the cortical sheet) different host species were used. After the staining procedure, sections greatly complicates the delineation of the parasubiculum. Here were mounted on gelatin-coated glass slides with Mowiol or Vectashield we aimed for a comprehensive description of parasubicular cir- mounting medium. In a subset of experiments, primary antibodies were cuits by a combined anatomical and functional approach (Bur- visualized by DAB staining. For this purpose, endogenous peroxidases were first blocked by incubating brain tissue sections in methanol con- galossi et al., 2011; Tang et al., 2014a). Specifically, we were taining 0.3% hydrogen peroxide in the dark at room temperature for 30 interested in how parasubicular circuits relate to pyramidal and min. The subsequent immunohistochemical procedures were performed stellate neuron microcircuits in layer 2 of medial entorhinal cor- as described above, with the exception that detection of primary antibod- tex (MEC) (Ray et al., 2014; Tang et al., 2014b). ies was performed by biotinylated secondary antibodies and the ABC In our current analysis, we investigate four issues: First, we detection kit. Immunoreactivity was visualized using DAB staining. delineate the location, shape, laminar organization, and internal The relative density of putative parvalbuminergic fibers in PV-Cre structure of parasubiculum. Second, we investigate the sources of mice in hippocampus CA1–3, presubiculum, parasubiculum, and medial parasubicular inputs, as well as the targets of parasubicular out- entorhinal cortex was estimated by manually outlining these four areas puts. Third, we assess spatial discharge patterns of parasubicular (Paxinos and Franklin, 2012) in epifluorescence images (2.5ϫ) from neurons by juxtacellular recording/labeling and tetrode record- horizontal sections (estimated depth 3 mm relative to bregma) and then ings in freely moving rats. Fourth, we assess the temporal dis- measuring mean fluorescence signals in each area with the ImageJ software. For comparison between brains, these values were then normalized charge patterns of identified and unidentified parasubicular to the mean hippocampal value in each brain (n ϭ 3). neurons, and how this might relate to anatomical connectivity. Anterograde and retrograde neuronal labeling. Anterograde or retrograde tracer solutions containing biotinylated dextrane amine (BDA) Materials and Methods (10% w/v; 3000 or 10,000 molecular weight) were injected in juvenile rats All experimental procedures were performed according to the Ger- (ϳ150 g) under ketamine/xylazine anesthesia. Briefly, a small craniot- man guidelines on animal welfare under the supervision of local eth- omy was opened above the parasubiculum/medial entorhinal cortex. ics committees. Before injection, the parasubiculum was localized by electrophysiologi- Brain tissue preparation. For anatomy experiments, male and female cal recordings, based on cortical depth, characteristic signatures of the Wistar rats (150–400 g) were anesthetized by isoflurane and then killed local field potential theta oscillations, and neuronal spiking activity. by an intraperitoneal injection of 20% urethane or sodium pentobarbital. Glass electrodes with a tip diameter of 10–20 ␮m, filled with BDA solu- They were then perfused transcardially with 0.9% PBS solution, followed tion, were then lowered into the target region. Tracers were either by 4% PFA in 0.1 M phosphate buffer (PB). After perfusion, brains were pressure-injected (10 injections using positive pressure of 20 psi, 10–15 s removed from the skull and postfixed in PFA overnight. They were then injection duration) or iontophoretically injected (7 s on/off current transferred into a 10% sucrose solution in PB and left overnight, and pulses of 1–5 mA for 15 min). After the injections, the pipettes were left in subsequently immersed in 30% sucrose solution for at least 24 h for place for several minutes and slowly retracted. The craniotomies were cryoprotection. The brains were embedded in Jung Tissue Freezing Me- closed by application of silicone and dental cement. The animals survived dium and subsequently mounted on the freezing microtome to obtain for 3–7 d before being transcardially perfused. 20- to 60-␮m-thick sagittal sections or tangential sections (parallel to the Viral injections and quantification of anterorgradely traced axons. pial surface of the MEC). Tangential sections were obtained by removing PV-Cre mice, expressing Cre recombinase under the PV promoter the cerebellum, visually identifying the pial surface of the MEC (Ray et (B6;129P2-Pvalbtm1(cre)Arbr/J mice, stock #008069, The Jackson al., 2014; their Fig. 1A), and making a cut 3 mm anterior and parallel Laboratory) were injected with AAV-Ef1a-dbf-hChR2(H134R)-EYFP- to the pial surface of the medial entorhinal cortex. The tissue was WPRE (serotype 1/2) ϳ6 weeks before perfusion. The medial septum was then frozen and positioned with the pial side to the block face of the targeted under stereotaxic guidance: starting from the pial surface at 1 microtome. mm anterior, 0.7 mm right lateral to bregma, a 34-gauge NanoFil needle Tissue from PV-Cre mice, expressing Cre recombinase under the (WPI) was advanced at an angle of 10° in the coronal plane for 4200 and parvalbumin (PV) promoter (B6;129P2-Pvalbtm1(cre)Arbr/J mice, 4600 ␮m, where we injected 1 ␮l each (100 nl/s), waiting 5 min after each stock #008069, The Jackson Laboratory), was prepared using similar injection before moving the needle. The AAV virus was generously pro- methods, except that the sections were cut on a standard microtome vided to us by Susanne Schoch (University of Bonn).

54 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2291

Fluorescence signals were normalized to dentate gyrus intensity levels Hz). This is referred to here as theta rhythmicity. Only neurons with and quantified. Briefly, regions of interests from horizontal sections (at a mean firing rate Ͼ0.5 Hz were included in the theta analysis. Statistical depth of ϳ3.5 mm ventral to bregma) (Paxinos and Franklin, 2012) were significance between groups was assessed by two-tailed Mann–Whitney manually outlined and the mean fluorescence intensity for each area nonparametric test with 95th confidence intervals. quantified using the ImageJ software (n ϭ 3 mice). Analysis of theta locking. For all neurons, we calculated the locking to Juxtacellular recordings. Juxtacellular recordings and tetrode record- theta phase based on spiking discharge in relation to theta rhythm in the ings in freely moving animals were obtained in male Wistar and Long– local field potential. The local field potential was zero phase bandpass Evans rats (150–250 g). Experimental procedures were performed as filtered (4–12 Hz), and a Hilbert transform was used to determine the recently described (Tang et al., 2014a, b, 2015). Briefly, rats were main- instantaneous phase of the theta wave. In line with previous studies (Miz- tained in a 12 h light/dark phase and were recorded in the dark phase. useki et al., 2009), the theta phase locking strength, S, and the preferred Glass pipettes with resistance 4–6 M⍀ were filled with extracellular Ring- phase angle, ␸, were defined as the modulus and argument of the Ray- er’s solution containing (in mM) the following: 135 NaCl, 5.4 KCl, 5 leigh average vector of the theta phase for all spikes. The theta phase

HEPES, 1.8 CaCl2, and 1 MgCl2, pH 7.2, and Neurobiotin (1%–2%). locking strength value can vary between 0 (uniform distribution of spikes Animal implantations were performed as previously described (Tang et over the theta cycle) and 1 (all spikes have the same theta phase). Only al., 2014a). To target the parasubiculum, a plastic ring was placed 0.2–0.5 spikes during running (speed cutoff ϭ 1 cm/s for juxtacellular signals, 5 mm anterior to the transverse sinus and 4.0–4.5 mm lateral to the mid- cm/s for tetrode recordings) were included in the analysis. Only neurons line. After implantation, rats were allowed to recover from the surgery with mean firing rate Ն0.5 Hz were included in the analysis. For com- and were habituated to head fixation for 3–5 d. Rats were trained in the parison to MEC L2 data, both the analysis procedures and the juxtacel- experimental arena (70 ϫ 70 cm or 1 ϫ 1 m square black box, with a lular dataset correspond to our recent publications (Ray et al., 2014; Tang white cue card on the wall) for 3–7 d. Juxtacellular recordings and label- et al., 2014b, 2015). ing were essentially performed as previously described (Tang et al., Analysis of spatial modulation. The position of the rat was defined as 2014a; Pinault, 1996). Unidentified recordings in parasubiculum were the midpoint between two head-mounted LEDs. A running speed either lost before the labeling could be attempted, or the recorded neu- threshold (see above) was applied for isolating periods of rest from active rons could not be unequivocally identified; but either pipette tracks or movement. Color-coded firing maps were plotted. For these, space was dendritic processes were found in the parasubiculum. After the experi- discretized into pixels of 2.5 cm ϫ 2.5 cm, for which the occupancy z of a ment, the animals were killed with an overdose of ketamine, urethane, or given pixel x was calculated as follows: sodium pentobarbital, and perfused transcardially with 0.1 M PB fol-

lowed by 4% PFA solution, shortly after the labeling protocol. The jux- z͑ x͒ ϭ w͉͑x Ϫ xt͉͒⌬t tacellular signals were amplified by the ELC-03XS amplifier (NPI ͸t Electronics) and sampled at 20 kHz by a data-acquisition interface under where x is the position of the rat at time t, ⌬t the interframe interval, and the control of PatchMaster 2.20 software (HEKA). The animal’s location t w a Gaussian smoothing kernel with ␴ ϭ 5 cm. and head-direction were automatically tracked at 25 Hz by the Neuralynx Then, the firing rate r was calculated as follows: video-tracking system and two head-mounted LEDs. MEC data for comparison have been published previously (Ray et al., 2014; Tang et al., 2014b, 2015). One head-direction cell recorded and identified in the ͸w͉͑x Ϫ xi͉͒ r͑ x͒ ϭ i parasubiculum has been shown in a previous paper (Tang et al., 2014a). z Tetrode recordings. Tetrode recordings from parasubiculum were essentially performed as recently described (Tang et al., 2014b, 2015). Te- where xi is the position of the rat when spike i was fired. The firing rate of trodes were turned from 12.5-␮m-diameter nichrome wire (California pixels, whose occupancy z was Ͻ20 ms, was considered unreliable and Fine Wire) and gold plated to ϳ250 k⍀ impedance. Spiking activity and not shown. local field potential were recorded at 32 kHz (Neuralynx; Digital Lynx). To determine the spatial periodicity of juxtacellularly recorded neu- The local field potential was recorded from the same tetrode as single rons, we determined spatial autocorrelations. The spatial autocorrelo- units and referenced to a superficial silent neocortical tetrode or to the rat gram was based on Pearson’s product moment correlation coefficient as ground. All recordings were performed following behavioral training, as follows: specified above for juxtacellular procedures. The animal’s location and head-direction were automatically tracked at 25 Hz by video tracking and n f͑ x, y͒ f͑ x Ϫ ␶ , y Ϫ ␶ ͒ head-mounted LEDs, as described above. After recordings, tetrodes were ͸ x y retracted from the parahippocampal areas, and multiple lesions were Ϫ f͑ x, y͒ f͑ x Ϫ ␶x, y Ϫ ␶y͒ performed at distinct sites along the individual tetrode tracks, thereby r͑␶ , ␶ ͒ ϭ ͸ ͸ allowing unequivocal assignment of the different tetrode tracks. Follow- x y 2 2 ing perfusion, brains were sectioned tangentially and recording sites as- ͱn ͸f͑ x, y͒ Ϫ ͩ͸f͑ x, y͒ͪ signed by histology. Spikes were preclustered using KlustaKwik (K.D. Harris, Rutgers University) and manually using MClust (A.D. Redish, 2 University of Minnesota). Cluster quality was assessed by spike shape, 2 ϫ n f͑ x Ϫ ␶x, y Ϫ ␶y͒ Ϫ f͑ x Ϫ ␶x, y Ϫ ␶y͒ ISI-histogram, L-ratio, and isolation distance. ͱ ͸ ͩ͸ ͪ Neurobiotin labeling and calbindin immunohistochemistry. For histological analysis of juxtacellularly labeled neurons, Neurobiotin was visu- where r͑␶x, ␶y͒ the autocorrelation between pixels or bins with spatial alized with streptavidin conjugated to Alexa-546 (1:1000). Subsequently, offset ␶x and ␶y.fis the image without smoothing or the firing rate map immunohistochemistry for Calbindin was performed, as previously de- after smoothing, n is the number of overlapping pixels or bins. Autocor- scribed (Ray et al., 2014), and visualized with AlexaFluor-488. After flu- relations were not estimated for lags of ␶x and ␶y, where n Ͻ 20. For orescence images were acquired, the Neurobiotin staining was converted spatial and head-directional analysis, both a spatial (Ͼ50% spatial cov- into a dark DAB reaction product. Neuronal morphologies were recon- erage) and a firing rate inclusion criterion (Ͼ0.5 Hz) were applied. Spa- structed by computer-assisted manual reconstructions (Neurolucida). tial coverage was defined as the fraction of visited pixels (bins) in the Analysis of theta rhythmicity. Theta rhythmicity of spiking discharge arena to the total pixels. was determined from the Fast Fourier Transform-based power spectrum Analysis of spatial information. For all neurons, we calculated the spa- of the spike-train autocorrelation functions of the neurons, binned at 10 tial information rate, I, from the spike train and rat trajectory as follows: ms. To measure modulation strength in the theta band (4–12 Hz), a theta power was computed, defined as the average power within 1 Hz of the 1 r͑x͒ maximum of the autocorrelation function in the theta rhythm (4–12 l ϭ r͑ x͒log2 o͑x͒dx T ͵ r៮

55 Ebbesen (2017) Chapter 5

2292 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

where r(x) and o(x) are the firing rate and occupancy as a function of a Results given pixel x in the rate map. r៮ is the overall mean firing rate of the cell, Geometry of the parasubiculum and T is the total duration of a recording session (Skaggs et al., 1993). A In our initial analysis, we sought to determine the general orga- cell was determined to have a significant amount of spatial information if nization of the parasubiculum. Tangential sections (parallel to the observed spatial information rate exceeded the 95th percentile of a distribution of values of I obtained by circular shuffling. Shuffling was the pial surface of the MEC; see Materials and Methods) of the performed by a circular time-shift of the recorded spike train relative to cortical sheet stained for acetylcholine esterase activity (Fig. 1A, the rat trajectory by a random time tЈ ʦ ͓0, T͔ for 1000 permutations left) or calbindin immunoreactivity (Fig. 1A, right) provide a (von Heimendahl et al., 2012; Bjerknes et al., 2014). particularly clear overview of the spatial extent of the parasubicu- Analysis of border cells. To determine the modulation of a cell firing lum. Consistent with findings from previous studies (Geneser, along a border, we determined border scores (Solstad et al., 2008). Border 1986; Slomianka and Geneser, 1991), we find that the parasu- fields were identified from a collection of neighboring pixels having a biculum shows prominent acetylcholine esterase activity (Fig. firing rate Ͼ0.3 times the maximum firing rate and covering an area of at 1A, left). The parasubiculum can also be identified by an absence least 100 cm (Sargolini et al., 2006). The coverage (Cm) along a wall was of calbindin immunoreactivity (Fig. 1A, left) (Fujise et al., 1995; defined as the maximum length of a putative border field parallel to a Boccara et al., 2010). Further subdivisions of the parasubiculum boundary, divided by the length of the boundary. The mean firing dis- have been suggested (Blackstad, 1956). Our data refer to the cal- tance (Dm) of a field was defined as the sum of the square of its distance bindin free area surrounding the MEC outlined in Figure 1A, left from the boundary, weighted by the firing rate (Solstad et al., 2008). The and highlighted in light blue in Figure 1B [possibly related to distance from a boundary was defined as the exponential of the square of the distance in pixels from the closest boundary, normalized by half “parasubiculum b” in the terminology of Blackstad (1956)]. Lat- the length of the boundary. Border scores were defined as the maxi- erally contiguous to the parasubiculum one observes a thin strip mum difference between Cm and Dm, divided by their sum, and of cortex containing numerous calbindin-positive neurons (Fig. ranged from Ϫ1 to 1. 1A, right, B, red “calbindin stripe”). Analysis of grid cells. Grid scores were calculated, as previously de- As shown in Figure 1A, B and quantified in Figure 1C, the scribed (Barry et al., 2012), by taking a circular sample of the autocorre- parasubiculum forms a fairly narrow (310 Ϯ 83 ␮m width, logram, centered on, but excluding the central peak. The Pearson N ϭ 10), but very elongated (5.190 Ϯ 0.485 mm length, N ϭ 10) correlation of this circle with its rotation for 60 degrees and 120 degrees continuous curved stripe, which flanks the medial entorhinal was obtained (on peak rotations) and also for rotations of 30, 90, and 150 cortex from its medial to dorsolateral side. The lateral part of the degrees (off peak rotations). Gridness was defined as the minimum dif- parasubiculum, dorsal to the medial entorhinal cortex, is nar- ference between the on-peak rotations and off-peak rotations. To deter- rower than the medial part. This may explain why this part of the mine the grid scores, gridness was evaluated for multiple circular samples parasubiculum has not been classified as such in most previous surrounding the center of the autocorrelogram with circle radii increas- studies (Boccara et al., 2010; Ding, 2013). Other histological ing in unitary steps from a minimum of 10 pixels more than the width of the radius of the central peak to the shortest edge of the autocorrelogram. markers, such as cytochrome-oxidase activity, or soma morphol- The radius of the central peak was defined as the distance from the central ogies, as visualized from Nissl stains (Burgalossi et al., 2011), also peak to its nearest local minima in the spatial autocorrelogram. The delineated the parasubiculum in the same way as shown in Figure radius of the inner circle was increased in unitary steps from the radius of 1 (data not shown). Similarly, parasagittal sectioning angles de- the central peak to 10 pixels less than the optimal outer radius. The grid lineate the same outlines of the parasubiculum. We conclude that score was defined as the best score from these successive samples. Grid the parasubiculum has a linear structure with a narrow width. scores reflect both the hexagonality in a spatial field and also the regular- We also investigated the laminar structure of the parasubicu- ity of the hexagon. To disentangle the effect of regularity from this index lum. Consistent with our previous conclusions (Burgalossi et al., and consider only hexagonality, we transformed the elliptically distorted 2011), we did not find direct evidence for a clear association of hexagon into a regular hexagon and computed the grid scores (Barry et deep layers with the parasubiculum. For example, following al., 2012). A linear affine transformation was applied to the elliptically tracer injections in the superficial parasubicular layers, we did not distorted hexagon, to stretch it along its minor axis, until it lay on a circle, observe back-labeled neurons in the adjacent deep layers, even with the diameter equal to the major axis of the elliptical hexagon. The grid scores were computed on this transformed regular hexagon (Barry et when we observed back-labeled neurons as distant as the subicu- al., 2012). lum (data not shown). Hence, we speculate that deep layers close Analysis of head-directionality. Head-direction tuning was measured as to the parasubiculum might not be part of this structure but the eccentricity of the circular distribution of firing rates. For this, firing could rather be associated with the neighboring medial entorhi- rate was binned as a function of head-direction (n ϭ 36 bins). A cell was nal cortex or the presubiculum (Mulders et al., 1997). said to have a significant head-direction tuning if the length of the average vector exceeded the 95th percentile of a distribution of average vector Internal structure of the parasubiculum lengths calculated from shuffled data and had a Rayleigh vector length Consistent with our previous observations (Burgalossi et al., Ͼ0.3. Data were shuffled by applying a random circular time-shift to the 2011), we found the superficial parts of the parasubiculum (cor- recorded spike train for 1000 permutations. responding to layers 1 and 2) can be divided into ϳ15 large Classification of cells into functional categories. Cells were classified as patches with a diameter ϳ500 ␮m each. These patches can be head-direction cells, grid cells, conjunctive cells, border cells, spatially revealed in superficial tangential sections (Fig. 1D, left) by PV irregular cells, and nonspatially modulated cells, based on their grid immunoreactivity and by cell density visualized by NeuN immu- score, border score, spatial information, and significance of head- noreactivity (Fig. 1D, right). However, the deeper parts of the directionality according to the following criteria: head-direction cells, parasubiculum (corresponding to layer 3) were not obviously Rayleigh vector length Ͼ0.3, and significant head-direction tuning (Boccara et al., 2010); grid cells, grid score Ͼ0.3, and significant spatial divided into patches (Fig. 1D). information; border cells, border score Ͼ0.5, and significant spatial in- Injections of the anterograde tracer BDA (3000 molecular formation (Solstad et al., 2008), or those who passed border test (Lever et weight) showed that parasubicular neurons extend long axons al., 2009); spatially irregular cells, significant spatial information throughout the full length of the parasubiculum (Fig. 1E), (Bjerknes et al., 2014), while not passing grid score or border score cri- consistent with previous evidence from single-cell microcircuits teria; and nonspatially modulated cell, no significant spatial information. (Burgalossi et al., 2011). In the latter work, these axons were

56 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2293

Figure 1. Shape and internal structure of the parasubiculum. A, Left, Tangential section stained for acetylcholinesterase activity (dark precipitate). The shape of the parasubiculum is outlined (white dashed line) coinciding with high acetylcholinesterase activity. Right, Tangential section (same section as in A, left) processed for calbindin immunoreactivity (green); the shape of the parasubiculumisnegativelyoutlinedbyanabsenceofcalbindinimmunoreactivity.B,Schematicoftheparasubiculum(lightblue)andadjacentMECsubdivisions.C,Quantificationofparasubiculum size in 10 hemispheres: length, width, and area. D, Tangential sections stained for PV immunoreactivity (green, left) and NeuN immunoreactivity (red, right). The parasubiculum stands out by its intensestaining.Threesectionsareshown:left,mostsuperficial(closesttothepia);middleandright,progressivelydeeper.Notehowthepatchystructureofthesuperficialparasubiculumisreplaced by a continuous cell band in deeper sections. E, Tangential sections of the parasubiculum showing the injection site of BDA tracer (red fluorescence) and anterogradely traced circumcurrent axons (accordingtotheterminologyofBurgalossietal.,2011),extendingthroughouttheparasubiculum(seealsomagnifiedinset,left).*Injectionsite.F,Parasagittalsectionsoftheparasubiculum(top) and parasubiculum and MEC (bottom) after the injection of larger amounts of BDA (tracer, dark color). The tracer completely fills the parasubiculum and stains layer 2 of the MEC. *Injection site. A, B, Modified from Ray et al. (2014). D, Dorsal; L, lateral; M, medial; V, ventral; A, anterior; P, posterior.

termed “circumcurrent,” as they appeared to interconnect para- these data point toward a strong medial septal drive to parasu- subicular patches. As a consequence of this internal connectivity, bicular neurons, likely contributing to strong theta rhythmicity a single tracer injection could label the full extent of the parasu- in the parasubiculum (Burgalossi et al., 2011; see below). biculum (Fig. 1F). This is a remarkable feature of the parasubicu- By retrograde-tracer injections, we also identified parasub- lum not seen in the medial entorhinal cortex. Thus, analysis of the iculum-projecting neurons in the anterior thalamus, subiculum, internal structure of parasubiculum indicates both modularity and presubiculum. The findings are consistent with the earlier con- and global connectivity. clusions of previous authors (Ko¨hler, 1985; van Groen and Wyss, 1992; Honda and Ishizuka, 2004) and are therefore not shown. Inputs to the parasubiculum Of particular interest for hippocampal function are the inputs Outputs from the parasubiculum from the medial septum, which are of critical importance to grid Previous work showed that the parasubicular axons innervate cell activity (Brandon et al., 2011; Koenig et al., 2011). We first layer 2 of the medial entorhinal cortex (van Groen and Wyss, sought to determine the patterns of GABAergic inputs from the 1990; Caballero-Bleda and Witter, 1993, 1994; but see Canto et medial septum, which are thought to play a critical role in theta- al., 2012). Recent work showed that principal neurons in layer 2 rhythm generation (Mitchell et al., 1982; Buzsa´ki, 2002; Hangya of medial entorhinal cortex segregate into stellate and pyramidal et al., 2009; Brandon et al., 2011; Koenig et al., 2011). To this end, cell subnetworks, which can be differentiated by the calbindin we performed viral injections in the medial septum in PV-Cre immunoreactivity of the pyramidal neurons (Varga et al., 2010). mice (see Materials and Methods) and expressed GFP selectively Layer 2 pyramidal neurons are arranged in a hexagonal grid, in GABAergic septal neurons (Fig. 2A,B). As shown in Figure 2A, show strong theta-rhythmic discharges (Ray et al., 2014), and the parasubiculum is an area within the hippocampal formation, might preferentially contribute to the grid cell population (Tang which receives a comparatively dense innervation from GABAe- et al., 2014b; but see Sun et al., 2015). To determine whether rgic medial septal neurons (as quantified by normalized fluores- parasubicular inputs target a specific subpopulation of neurons cence levels, mean values [n ϭ 3], parasubiculum ϭ 1.5 Ϯ 0.23, in layer 2 of medial entorhinal cortex, we performed fine-scale MEC ϭ 1.0 Ϯ 0.05, CA1–3 ϭ 1.2 Ϯ 0.06, PreS ϭ 1.0 Ϯ 0.04; Fig. injections of anterograde tracers in the dorsal parasubiculum, 2C). As we already noted earlier, there is also a prominent expres- combined with visualization of calbindin patterns (Fig. 3). As sion of cholinergic activity markers (Figs. 1A, 2D) in line with in shown in Figure 3, tangential sections through layer 2 with cal- vitro work showing robust response of parasubicular neurons to bindin immunostaining revealed a regular organization of muscarinic activation (Glasgow and Chapman, 2013). Together, patches of pyramidal neurons (Ray et al., 2014). Surprisingly,

57 Ebbesen (2017) Chapter 5

2294 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

Figure 2. Parasubiculum receives GABAergic and cholinergic inputs. A, Horizontal sections showing that the parasubiculum contains the densest projection in the hippocampal formation of GFP-positive, putative parvalbuminergic fibers deriving from injection of AAV into the medial septum (inset, asterisk) of mice expressing Cre recombinase under the PV promoter. This dense projection pattern was seen in 3 of 3 injected mice. In this brain, olfactory and accessory olfactory areas were also labeled unilaterally. B, PV immunostaining (red) marks the extent of the parasubiculum(asshowninFig.1D,left).NotethehigherdensityofGABAergicmedialseptalfibers(green)withintheparasubiculum.C,Normalizedfluorescenceintensitylevelsrelativetodentate gyrus (n ϭ 3 mice, different gray symbols represent the different mice). Green squares represent mean normalized fluorescence. D, Tangential section showing high levels of acetylcholinesterase in the parasubiculum. PrS, Presubiculum; DG, dentate gyrus; LEC, lateral entorhinal cortex; PaS, parasubiculum; Por, postrhinal cortex; Per, perirhinal cortex; CA, cornus ammonis.

these patches were selectively innervated by parasubicular af- In line with previous observations in linear mazes (Burgalossi ferents (Fig. 3A,B), which targeted the center of patches (Fig. et al., 2011), many juxtacellularly recorded neurons showed 3C). This indicates that parasubicular axons may preferen- head-direction selectivity. A representative neuron is shown in tially target layer 2 pyramidal neurons of medial entorhinal Figure 4E. This neuron was situated in the medial part of the cortex, which may in turn contribute to the strong theta rhyth- parasubiculum (Fig. 4E, right) and also discharged in bursts with micity in these neurons (Ray et al., 2014). strong theta rhythmicity (Fig. 4F, right). The spiking autocorrelogram also revealed a strong theta rhythmicity (Fig. 4G, left), and Identification of functional cell types in the parasubiculum the spikes were strongly locked to local theta oscillations (Fig. 4G, Compared with its major target structure (the entorhinal cortex), right). Spikes were fired throughout the enclosure without obvi- limited information is currently available about the spatial dis- ous spatial modulation (Fig. 4H, left) but showed a clear head- charge properties in the parasubiculum (Taube, 1995; Cacucci et direction preference (Fig. 4H, right). al., 2004; Boccara et al., 2010). To address this issue, we juxtacellularly recorded and labeled neurons (n ϭ 16) in the parasubiculum of freely moving rats trained to explore 2D environments Spatial firing properties of parasubicular neurons (Tang et al., 2014a) A representative recording from an identified By combining juxtacellularly recorded and identified parasu- parasubicular neuron is shown in Figure 4A. This neuron had bicular neurons with verified recording sites of single-cell and divergent sideward-directed dendrites (seen from the top), was tetrode recordings (see Materials and Methods), we could pro- situated in the dorsal part of the parasubiculum (Fig. 4A, right), vide a more comprehensive characterization of functional cell and discharged in spike bursts strongly entrained by the theta types in parasubiculum. In line with previous work (Boccara et rhythm (Fig. 4B). Theta rhythmicity of spiking was revealed by al., 2010), we observed border discharges (9%, 6 of 68; Figs. 4D, the spiking autocorrelogram (Fig. 4C, left), and the spikes were 5A), grid discharges (9%, 6 of 68; Fig. 5B), strong head-direction also strongly locked to local theta oscillations (Fig. 4C, right). The selectivity (50%, 34 of 68; Figs. 4H, 5C), and a substantial pro- neuron discharged along the border of the enclosure (Fig. 4D, portion of irregular spatial discharges (40%, 27 of 68) (cells not left), a defining feature of border activity (Solstad et al., 2008), shown). This last group contains cells with significant spatial and showed head-direction selectivity (Fig. 4D, right). information content (Skaggs et al., 1993; see Materials and Meth-

58 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2295

tended to have a higher theta rhythmicity than identified layer 2 neurons (juxtacellularly recorded cells, p ϭ 0.0116, Mann– Whitney U test), but this difference did not reach statistical significance when tetrode units were included in the sample of parasubicular neurons. Theta phase locking strength (mean (circular) vector length; see Materials and Methods) of parasubicular neurons was similar to that of MEC layer 2 pyramidal neurons (p Ͼ 0.05, Mann–Whitney U test: z ϭϪ0.89; Fig. 7D) and significantly stronger than that of layer 2 stellates and layer 3 neurons (both p Ͻ 0.001, Mann–Whitney U test: z(Stel) ϭ 3.73, z(L3) ϭ 7.83; Fig. 7D; MEC cells from Tang et al., 2014b, 2015). Notably, at the population level, parasubicular and MEC layer 2 pyramidal and stellate neurons showed distinct preferred theta phases (all p Ͻ 0.05, Rayleigh test for nonuniformity: z(parasubiculum) ϭ 29.5, z(Pyr) ϭ 4.07, z(Stel) ϭ 3.36; Fig. 7E; MEC cells from Tang et al., Figure 3. Parasubicular axons target layer 2 pyramidal cell patches in medial entorhinal cortex. A, Left, Tangential section 2014b, 2015). When we compared the stained for calbindin (green) revealing patches of calbindin-positive pyramidal neurons. Middle, Same section as left processed to preferred phase of identified MEC layer 2 reveal the tracer BDA (red). *Location of the parasubicular injection site. Right, Overlay. Scale bar, 150 ␮m. B, Same as A but at higher magnification. Scale bar, 150 ␮m. C, High-magnification view of a single patch. D, Dorsal; L, lateral; M, medial; V, ventral. pyramids and identified parasubicular neurons, we found that the parasubicular neurons preferred an earlier theta phase ods) but that do not meet grid or border inclusion criteria (see (slightly before the trough; Fig. 7F, left; p ϭ 0.048, Watson–Wil- Materials and Methods). liams test for equal circular means: F ϭ 4.34). When we included Next, we compared the spatial discharge properties of the all nonidentified juxta and tetrode recordings of parasubicular parasubiculum with those of identified and putative MEC layer 2 and putative MEC layer 2 pyramidal neurons, this difference re- pyramidal and stellate neurons (Tang et al., 2014b) as well as mained statistically significant (Fig. 7F, right; 155° vs 174°, p ϭ neurons recorded in MEC layer 3 (Tang et al., 2015). We found 0.0000085, Watson–Williams test for equal circular means: F ϭ significantly more spatial responses in the parasubicular neurons 22.7). Because tetrode recordings of MEC layer 2 were assigned than in the other cell types (Fig. 6A; all p Ͻ 0.01, ␹ 2 test with their putative cell identity based on their temporal spiking prop- Bonferroni–Holm correction: 39 of 68 parasubiculum, ␹ 2 ϭ 7.91 erties, we wondered whether this might have biased the compar- vs 35 of 99 Pyr, ␹ 2 ϭ 12.4 vs 28 of 94 Stel, ␹ 2 ϭ 11.1 vs 19/66 L3). ison of preferred theta phase. However, two indications suggest We also observed a strong head-directionality of parasubicular that this was not the case: (1) in the identified dataset, we had not neurons, in line with previous observations from linear track excluded any MEC layer 2 neurons, which locked before the recordings (Burgalossi et al., 2011). At the population level, the trough (Tang et al., 2014b); and (2) even when we applied the median head-direction vector of all parasubicular neurons was same classifier to all parasubicular neurons and only compared 0.31, much larger than in MEC layer 2 (0.12 in pyramidals; 0.14 in MEC layer 2 putative pyramids with parasubicular neurons, stellates) and layer 3 (0.09 in layer 3 cells; Fig. 6B; all p Ͻ 0.001, which would have been classified as putative pyramids, the Mann–Whitney U tests with Bonferroni–Holm correction: difference in preferred phase was still trending toward signif- z(Pyr) ϭ 5.54, z(Stel) ϭ 5.79, z(L3) ϭ 7.01). Similarly, the pro- icance (p ϭ 0.076, Watson–Williams test for equal circular portion of neurons classified as head-direction cells was also con- means: F ϭ 3.21). siderably larger than in MEC layers 2 and 3 (Fig. 6C; all p Ͻ 0.001, The strong theta phase locking strength and theta rhythmicity ␹ 2 test with Bonferroni–Holm correction: ␹ 2(Pyr) ϭ 17.7, of both parasubicular neurons and layer 2 pyramidal (but not ␹ 2(Stel) ϭ 28.8, ␹ 2(L3) ϭ 30.7) stellate) neurons, as well as the preference of parasubicular neurons to fire at a slightly earlier theta phase (ϳ19° phase angle, i.e., Theta modulation of parasubicular neurons ϳ7 ms, assuming an 8 Hz theta rhythm) than layer 2 pyramidal As shown in representative neurons (Figs. 4, 5), the large majority of neurons, are consistent with the idea that parasubicular neurons parasubicular neurons showed strong theta rhythmicity, as revealed might impose a feedforward theta-modulated drive onto layer 2 by autocorrelation of spike trains (Fig. 7A)(Tang et al., 2014b,2015). pyramidal neurons. Parasubicular neurons were also strongly locked to local field potential theta oscillations, which is known to be in phase with MEC theta Discussion (Glasgow and Chapman, 2007)(Fig. 7B). On average, theta rhyth- Unique features of the parasubiculum micity was stronger in parasubicular neurons than in identified layer The parasubiculum is distinct from other parahippocampal struc- 2 stellates and layer 3 neurons (both p Ͻ 0.01, Mann–Whitney U tures. The elongated shape of the parasubiculum and an almost lin- test, z(Stel) ϭ 3.19, z(L3) ϭ 8.39; Fig. 7C; MEC cells from Tang et al., ear arrangement of neurons differ from other (para)hippocampal 2014b and Tang et al., 2015). Identified parasubicular neurons structures, such as dentate gyrus, CA3, CA2, CA1, subiculum, pre-

59 Ebbesen (2017) Chapter 5

2296 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

Figure4. Physiologyofidentifiedparasubicularneurons.A,Left,Reconstructionofabordercelljuxtacellularlyrecordedandidentifiedinaratexploringa2Denvironment(70ϫ70cm).Redand blue represent reconstructed dendrites and axon, respectively. Scale bar, 100 ␮m. Right, Schematic of the location of the cell in the parasubiculum (arrow). The cell is located in the dorsal band of the parasubiculum (blue), close to medial entorhinal cortex (gray). Scale bar, 1000 ␮m. B, Representative raw traces of the recorded cell shown in A. Note the prominent theta rhythm in LFP and theta-modulated firing of the recorded cell. C, Left, Autocorrelogram of spike discharges for the cell shown in A. Right, theta phase histogram of spikes for the cell shown in A. For convenience, two repeated cycles are shown. The black sinusoid is a schematic local field potential theta wave for reference. D, Spike-trajectory plot (left) and rate map (middle) revealing the border firing. Spike-trajectory plot, Red dots indicate spike locations. Gray lines indicate the rat trajectory. Rate map, Red represents maximal firing rate, value noted above. For this cell, the border score is 0.86. Right, Polar plot of the cell’s head-direction tuning. Value indicates maximum firing rate to the preferred direction. E–H, Same as A–D for an identified head-direction cell. D, Dorsal; L, lateral; M, medial; V, ventral.

subiculum, and medial or lateral entorhinal cortex (Amaral and torhinal cortex and the presubiculum (Amaral and Witter, 1989; Witter, 1989; Cenquizca and Swanson, 2007). Further, absence of Cenquizca and Swanson, 2007). We provide evidence that the directly associated deep layers distinguishes the parasubiculum from parasubiculum extends further laterally than previously thought the surrounding entorhinal, retrosplenial, and presubicular cortices. (van Strien et al., 2009; Boccara et al., 2010) and that this struc- The “circumcurrent” axons (as defined by Burgalossi et al., 2011) ture might lack direct association with deep layers. The idea that (Fig. 1E,F) that traverse the parasubiculum and could thus establish the parasubiculum extends dorsolaterally from the medial ento- a “global” connectivity are also a unique feature of parasubicular rhinal cortex is based on three observations: (1) staining of cho- anatomy. Furthermore, the parasubiculum is a preferred target of linergic markers, calbindin immunoreactivity, or cytochrome medial septal inputs and provides the major input to pyramidal neu- oxidase activity all delineate a continuous band, which extends ron patches in layer 2 of medial entorhinal cortex. We observed a dorsolaterally; similarly, both (2) the modular structure of the larger fraction of spatial and head-directional responses in the para- “large patches” and (3) “circumcurrent” axons extend as a con- subiculum than in the adjacent medial entorhinal cortex (Solstad et tinuous dorsolateral band (Fig. 1)(Burgalossi et al., 2011). Our al., 2008; Tang et al., 2014b). conclusion that the parasubiculum extends dorsolaterally is strongly supported by recent high-resolution mapping of gene Comparison with previous work expression in parahippocampal cortices (Ramsden et al., 2015). Our anatomical analysis agrees with earlier descriptions that large The authors not only observed that this dorsolateral part is dif- parts of the parasubiculum are situated between the medial en- ferent from medial entorhinal cortex but also showed that it

60 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2297

Figure 5. Border, grid, and head-direction firing properties of parasubicular neurons. A, Parasubicular neurons classified as border cells. Left to right, Spike-trajectory plot, rate map, 2D spatial autocorrelation, angular tuning (shown only for head-direction selective cells), and spike autocorrelogram. Numbers above the rate map indicate maximum (Figure legend continues.)

61 Ebbesen (2017) Chapter 5

2298 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

shares patterns of gene expression with the “classical” medial parasubiculum (Ramsden et al., 2015). The extent to which deep layers were assigned to the parasubiculum varies in the literature. Whereas some studies assigned deep layers to the parasubiculum (Funahashi and Stewart, 1997; Glasgow and Chapman, 2007; Boccara et al., 2010), other work found it difficult to assign adjacent deep layers to either the presubiculum or the parasubiculum based solely on cytoarchitectonic criteria (Mulders et al., 1997). Our assessment that these deep layer neurons should not be viewed as part of the parasubiculum is based on three observations: (1) the shape of dorsal part of the parasubiculum, as revealed by cholinergic markers, calbindin immunoreactivity, or cytochrome oxidase activity, delineates only a “superficial-layer structure” encompassing layers 1–3 (Burgalossi et al., 2011); (2) we did not observe axons from the superficial parasubiculum into adjacent deep cortical layers; and (3) we did not observe axons from the adjacent deep cortical layers into the superficial parasubiculum. The idea that large parts of the parasubiculum lack deep layers is again supported by the gene expression analysis of Ramsden et al. (2015). Our results agree with previous extracellular recording data that also revealed the presence of spatially modulated neurons in the parasubiculum (Taube, 1995; Cacucci et al., 2004; Boccara et al., 2010; Burgalossi et al., 2011). The present data are also consistent with the study of Boccara et al. (2010), where the authors described grid, border, and head-direction responses in the parasubiculum. Notably, the strong head-direction tuning in the parasubiculum is also consistent with previous (Fyhn et al., 2008;Wills et al., 2010) and more recent work (Giocomo et al., 2014), where sharply tuned head- direction neurons were recorded “near” the dorsalmost border medial entorhinal cortex, hence compatible with a parasubicular origin of these signals (Fig. 1)(Burgalossi et al., 2011). Extracellular recordings have also identified both theta-rhythmic and non–theta- rhythmic border cells in this dorsalmost region of MEC (Solstad et al., 2008), where the parasubiculum extends in a narrow stripe above MEC (Fig. 1A). We found that parasubicular border cells lock strongly to the theta rhythm, whereas border cells in MEC layer 2 show only weak entrainment by theta oscillations (Tang et al., 2014b). Our results show a substantial proportion of spatially irregular cells, in line with previous work (Krupic et al., 2012), which also showed a larger percentage of nongrid, spatially modulated cells in the parasubiculum compared with adjacent MEC. Spatially irregular cells could provide sufficient spatial information for coding the animal’s position in space (Zhang et al., 1998; Zhang and Sejnowski, 1999).

Parasubicular discharge properties mirror those of its input structures Parasubicular response properties match well with the properties of its inputs. Parasubicular head-direction selectivity is in line with its inputs from anterior thalamus and presubiculum (Taube, 2007). The border responses observed here are in line with subicular inputs, as numerous boundary-vector cells have Figure 6. Border and head-direction (HD) firing properties of parasubicular neurons. Data from been observed there (Lever et al., 2009). A prominent aspect of layers2and3ofMECcomefromtheworkofTangetal.(2014b,2015)andareshownforcomparison. A,Comparisonoffractionsofspatialdischargesforparasubiculum,MECL2pyramidal,MECL2stellate, and MEC L3 neurons. Parasubicular neurons show large fraction of significantly spatially modulated 4 cells:gridcells,bordercells,andspatiallyirregularcells(␹2testwithBonferroni–Holmcorrection).B, (Figure legend continued.) firing rate. Numbers above the angular tuning map indicate max- ComparisonofHDvectorlengthsforparasubiculum(blue),MECL2pyramidal(green),MECL2stellate imum firing rate at the preferred direction. Scale bar (below the spike trajectory plot), 50 cm. (black), and MEC L3 (red) neurons. Parasubicular neurons show significantly higher average HD vector *Border cell recorded in a 70 ϫ 70 cm arena. Preborder and Postborder refer to the border test length than all others (Mann–Whitney Utests with Bonferroni–Holm correction). Lines indicate medians. (recording of the same cell before and after the introduction of an additional wall into the Horizontaldottedlineat0.3indicatesthethresholdforHDclassification.C,ComparisonoffractionsofHDcells arena). B, Parasubicular neurons classified as grid cells (same panels as in A). C, Parasubicular for parasubiculum, MEC L2 pyramidal, MEC L2 stellate, and MEC L3 neurons. Parasubicular neurons show 2 neurons classified as head-direction cells (see Materials and Methods). Left, Angular tuning. significantlyhigherpercentageofHDcells(␹ testwithBonferroni–Holmcorrection). Right, Spike autocorrelogram. Conventions as in A.

62 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2299

Figure 7. Theta modulation of parasubicular neurons compared with superficial medial entorhinal cortex. Data from layers 2 and 3 of MEC come from the work of Tang et al. (2014b, 2015) and are shown here for comparison. A, Representative autocorrelograms of spike discharges of identified neurons recorded from parasubiculum (blue), MEC L2 pyramidal (L2P, green), MEC L2 stellate (L2S, black) and MEC L3 (red) neurons. B, Theta phase histogram of spikes for the neurons shown in A. For convenience, two theta cycles are shown. The black sinusoid is a schematic local field potentialthetawaveforreference.C,Comparisonofthepowerofthetarhythmicityinparasubiculum(blue),MECL2pyramid(L2P,green),MECL2stellate(L2S,black),andMECL3(L3,red)neurons. Parasubiculum neurons show significantly stronger theta rhythmicity than MEC L2 stellate and MEC L3 neurons (Kruskal–Wallis test with Bonferroni correction). Lines indicate medians. D, Comparison of the theta phase locking strength (abbreviated in the figure as “Locking strength”) for the neurons shown in C. Parasubicular and MEC L2 pyramidal neurons show significantly higher theta phase locking than MEC L2 stellate and MEC L3 neurons (Kruskal–Wallis test with Bonferroni–Holm correction; significant differences between L2P and L2S have been shown in Ray et al., 2014). Lines indicate medians. E, Comparison of the preferred theta phase for the neurons shown in C. Parasubicular and MEC L2 neurons show significant preferred theta phases, whereas MEC L3 neuronsdonot(RayleightestfornonuniformitywithBonferroni–Holmcorrection:coloredpvaluesontherightside;Watson–WilliamstestforequalmeanswithBonferroni–Holmcorrection:black linedpvalues,top).Coloredlinesindicatecircularmeans.F,Polarplotsofpreferredthetaphase(thetapeakϭ0°)andthetaphaselockingstrength(Rayleighvector,0–1)forparasubiculum(blue) and MEC L2 pyramidal (green). Left, Only identified neurons. Right, Identified neurons and tetrode recordings. Dots represent identified neurons. Triangles represent tetrode units. Line indicates mean direction, median strength of locking.

parasubicular activity is the strong theta phase locking and theta states promoting theta oscillations (Glasgow and Chapman, rhythmicity of spike discharges. Large membrane-potential theta 2007, 2013). oscillations have also been recorded from parasubicular neurons in awake animals (Domnisoru et al., 2013). Such strong entrain- Does the parasubiculum provide input to the grid system? ment may result from the massive septal GABAergic innervation Our data suggest a relationship between the parasubiculum and (Fig. 3A,B) because GABAergic neurons in the medial septum layer 2 of medial entorhinal cortex grid cells (Sargolini et al., are known to be a key theta pacemaker (Buzsa´ki, 2002; Hangya et 2006; Boccara et al., 2010). Grid cells in the medial entorhinal al., 2009; Brandon et al., 2011; Koenig et al., 2011). Cholinergic cortex show strong theta rhythmicity of spiking (Boccara et al., innervation may also drive parasubicular neurons to depolarized 2010). It is therefore most interesting that the strongly theta-

63 Ebbesen (2017) Chapter 5

2300 • J. Neurosci., February 17, 2016 • 36(7):2289–2301 Tang et al. • Functional Architecture of the Rat Parasubiculum

rhythmic parasubicular neurons project selectively into layer 2 hippocampal formation: a review of anatomical data. Neuroscience 31: pyramidal cell patches, where neurons show strong entrainment 571–591. CrossRef Medline by the theta rhythm (Ray et al., 2014) and where most grid cells Andersen P, Bliss TV, Skrede KK, Lomo T, Olsen LI (1971) Lamellar organization of hippocampal excitatory pathways. Exp Brain Res 13:222–238. might be located (Tang et al., 2014b). The discharge timing is Medline consistent with an activation/entrainment of layer 2 pyramidal Barry C, Ginzberg LL, O’Keefe J, Burgess N (2012) Grid cell firing patterns neurons by parasubicular inputs. Parasubicular neurons dis- signal environmental novelty by expansion. Proc Natl Acad Sci U S A charge on average at an earlier theta phase (ϳ19° phase angle, i.e., 109:17687–17692. CrossRef Medline ϳ7 ms) than layer 2 pyramidal neurons (Fig. 7E,F). The parasu- Bjerknes TL, Moser EI, Moser MB (2014) Representation of geometric bor- bicular input to layer 2 pyramidal neurons is also remarkable, in ders in the developing rat. Neuron 82:71–78. CrossRef Medline light of the sparse excitatory connectivity within layer 2 of medial Blackstad TW (1956) Commissural connections of the hippocampal region in the rat, with special reference to their mode of termination. J Comp entorhinal cortex (Couey et al., 2013; Pastoll et al., 2013). Para- Neurol 105:417–537. CrossRef Medline subicular inputs could be important for three aspects: (1) for Boccara CN, Sargolini F, Thoresen VH, Solstad T, Witter MP, Moser EI, imposing theta rhythmicity on grid responses, and possibly also Moser MB (2010) Grid cells in pre- and parasubiculum. Nat Neurosci contributing to their temporal spiking dynamics (Hafting et al., 13:987–994. CrossRef Medline 2008; Mizuseki et al., 2009; Ray et al., 2014; Tang et al., 2014b). Bonnevie T, Dunn B, Fyhn M, Hafting T, Derdikman D, Kubie JL, Roudi Y, (2) Parasubicular head-directional responses could be causally Moser EI, Moser MB (2013) Grid cells require excitatory drive from the related to downstream grid activity in layer 2 of MEC. Indeed, hippocampus. Nat Neurosci 16:309–317. CrossRef Medline Brandon MP, Bogaard AR, Libby CP, Connerney MA, Gupta K, Hasselmo grid cells have been shown to receive head-directional inputs, and ME (2011) Reduction of theta rhythm dissociates grid cell spatial peri- disruptions of head-direction signals also impaired grid cell firing odicity from directional tuning. Science 332:595–599. CrossRef Medline (Bonnevie et al., 2013; Winter et al., 2015). The parasubiculum Brun VH, Solstad T, Kjelstrup KB, Fyhn M, Witter MP, Moser EI, Moser MB might be the source of this input, given the large fraction of (2008) Progressive increase in grid scale from dorsal to ventral medial head-direction cells and the selective output to MEC layer 2 (Fig. entorhinal cortex. Hippocampus 18:1200–1212. CrossRef Medline 3). (3) Parasubicular border activity could be needed for anchor- Burgalossi A, Herfst L, von Heimendahl M, Fo¨rste H, Haskic K, Schmidt M, ing entorhinal layer 2 grids to environmental boundaries (Hard- Brecht M (2011) Microcircuits of functionally identified neurons in the rat medial entorhinal cortex. Neuron 70:773–786. CrossRef Medline castle et al., 2015). Interestingly, direct projections from border Burgess N (2014) The 2014 Nobel Prize in Physiology or Medicine: a spatial to grid cells have been recently postulated, which might be re- model for cognitive neuroscience. Neuron 84:1120–1125. CrossRef sponsible for determining grid orientation, ellipticity, and stabil- Medline ity (Kruge et al., 2014; Hardcastle et al., 2015; Krupic et al., 2015; Buzsa´ki G (2002) Theta oscillations in the hippocampus. Neuron 33: Stensola et al., 2015). The parasubiculum might be one source of 325–340. CrossRef Medline border signals into the entorhinal grid system. Caballero-Bleda M, Witter MP (1993) Regional and laminar organization of projections from the presubiculum and parasubiculum to the entorhinal cortex: an anterograde tracing study in the rat. J Comp Neurol 328: Functional considerations 115–129. CrossRef Medline What does the parasubiculum do? It seems likely that the para- Caballero-Bleda M, Witter MP (1994) Projections from the presubiculum subiculum plays a role in determining spike timing of down- and the parasubiculum to morphologically characterized entorhinal- stream neurons relative to theta oscillations. The massive internal hippocampal projection neurons in the rat. Exp Brain Res 101:93–108. connectivity of the parasubiculum by circumcurrent axons is Medline rather unique. These axons connect along the dorsoventral axis Cacucci F, Lever C, Wills TJ, Burgess N, O’Keefe J (2004) Theta-modulated of the parahippocampal cortex. As different spatial scales are place-by-direction cells in the hippocampal formation in the rat. J Neu- rosci 24:8265–8277. CrossRef Medline mapped onto the dorsoventral axis of the medial entorhinal cor- Canto CB, Koganezawa N, Beed P, Moser EI, Witter MP (2012) All layers of tex (Brun et al., 2008), we wonder whether these axons ensure medial entorhinal cortex receive presubicular and parasubicular inputs. that those parasubicular neurons along the dorsoventral axis sig- J Neurosci 32:17620–17631. CrossRef Medline naling the same positions (at different spatial scales) fire at the Cenquizca LA, Swanson LW (2007) Spatial organization of direct hip- same time relative to the theta cycle. Another peculiar aspect of pocampal field CA1 axonal projections to the rest of the cerebral cortex. parasubicular anatomy is the lack of strong direct hippocampal Brain Res Rev 56:1–26. CrossRef Medline connections (van Strien et al., 2009). Together with the absence Couey JJ, Witoelar A, Zhang SJ, Zheng K, Ye J, Dunn B, Czajkowski R, Moser MB, Moser EI, Roudi Y, Witter MP (2013) Recurrent inhibitory cir- of deep layers (the recipient of CA1/subicular back projections in cuitry as a mechanism for grid formation. Nat Neurosci 16:318–324. the medial entorhinal cortex) and a thinner layer 1, it seems that CrossRef Medline the parasubiculum is only poorly connected to the “trisynaptic Ding SL (2013) Comparative anatomy of the prosubiculum, subiculum, memory loop” (reciprocal connections between the parasubicu- presubiculum, postsubiculum, and parasubiculum in human, monkey, lum and postrhinal cortex could, however, provide an indirect and rodent. J Comp Neurol 521:4145–4162. CrossRef Medline pathway) (Agster and Burwell, 2013). We envision that the para- Domnisoru C, Kinkhabwala AA, Tank DW (2013) Membrane potential dy- subiculum may function more for providing online spatial infor- namics of grid cells. Nature 495:199–204. CrossRef Medline Fujise N, Hunziker W, Heizmann CW, Kosaka T (1995) Distribution of the mation like a pointer (“where am I?”) rather than for long-term calcium binding proteins, calbindin D-28K and parvalbumin, in the storage of information (“where was I?”). This pointer hypothesis subicular complex of the adult mouse. Neurosci Res 22:89–107. CrossRef is consistent with disruption of place cell activity (Liu et al., 2004) Medline and working memory deficits after parasubicular lesions (Kesner Funahashi M, Stewart M (1997) Presubicular and parasubicular cortical and Giles, 1998). neurons of the rat: functional separation of deep and superficial neurons in vitro. J Physiol 501:387–403. CrossRef Medline Fyhn M, Hafting T, Witter MP, Moser EI, Moser MB (2008) Grid cells in References mice. Hippocampus 18:1230–1238. CrossRef Medline Agster KL, Burwell RD (2013) Hippocampal and subicular efferents and Geneser FA (1986) Distribution of acetylcholinesterase in the hippocampal afferents of the perirhinal, postrhinal, and entorhinal cortices of the rat. region of the rabbit: I. Entorhinal area, parasubiculum, and presubicu- Behav Brain Res 254:50–64. CrossRef Medline lum. J Comp Neurol 254:352–368. CrossRef Medline Amaral DG, Witter MP (1989) The three-dimensional organization of the Giocomo LM, Stensola T, Bonnevie T, Van Cauter T, Moser MB, Moser EI

64 Ebbesen (2017) Functional Architecture of the Parasubiculum

Tang et al. • Functional Architecture of the Rat Parasubiculum J. Neurosci., February 17, 2016 • 36(7):2289–2301 • 2301

(2014) Topography of head-direction cells in medial entorhinal cortex. Ramsden HL, Su¨rmeli G, McDonagh SG, Nolan MF (2015) Laminar and Curr Biol 24:252–262. CrossRef Medline dorsoventral molecular organization of the medial entorhinal cortex re- Glasgow SD, Chapman CA (2007) Local generation of theta-frequency EEG vealed by large-scale anatomical analysis of gene expression. PLoS Com- activity in the parasubiculum. J Neurophysiol 97:3868–3879. CrossRef put Biol 11:e1004032. CrossRef Medline Medline Ray S, Naumann R, Burgalossi A, Tang Q, Schmidt H, Brecht M (2014) Glasgow SD, Chapman CA (2013) Muscarinic depolarization of layer II Grid-layout and theta-modulation of layer 2 pyramidal neurons in medial neurons of the parasubiculum. PLoS One 8:e58901. CrossRef Medline entorhinal cortex. Science 343:891–896. CrossRef Medline Hafting T, Fyhn M, Bonnevie T, Moser MB, Moser EI (2008) Hipp- Sargolini F, Fyhn M, Hafting T, McNaughton BL, Witter MP, Moser MB, ocampus-independent phase precession in entorhinal grid cells. Nature Moser EI (2006) Conjunctive representation of position, direction, and 453:1248–1252. CrossRef Medline velocity in entorhinal cortex. Science 312:758–762. CrossRef Medline Hangya B, Borhegyi Z, Szila´gyi N, Freund TF, Varga V (2009) GABAergic Skaggs WE, McNaughton BL, Gothard KM, Markus EJ (1993) An neurons of the medial septum lead the hippocampal network during theta information-theoretic approach to deciphering the hippocampal code. activity. J Neurosci 29:8094–8102. CrossRef Medline In: Advances in neural processing systems, Vol. 5 (Hanson SJ, Cowan JD, Hardcastle K, Ganguli S, Giocomo LM (2015) Environmental boundaries as Giles CL, eds). San Mateo, CA: Morgan Kaufmann. an error correction mechanism for grid cells. Neuron 86:827–839. Slomianka L, Geneser FA (1991) Distribution of acetylcholinesterase in the CrossRef Medline hippocampal region of the mouse: I. Entorhinal area, parasubiculum, Honda Y, Ishizuka N (2004) Organization of connectivity of the rat pre- retrosplenial area, and presubiculum. J Comp Neurol 303:339–354. subiculum: I. Efferent projections to the medial entorhinal cortex. CrossRef Medline J Comp Neurol 473:463–484. CrossRef Medline Solstad T, Boccara CN, Kropff E, Moser MB, Moser EI (2008) Representa- Kesner RP, Giles R (1998) Neural circuit analysis of spatial working mem- tion of geometric borders in the entorhinal cortex. Science 322: ory: role of pre- and parasubiculum, medial and lateral entorhinal cortex. 1865–1868. CrossRef Medline Hippocampus 8:416–423. CrossRef Medline Stensola T, Stensola H, Moser MB, Moser EI (2015) Shearing-induced asymmetry Koenig J, Linder AN, Leutgeb JK, Leutgeb S (2011) The spatial periodicity of in entorhinal grid cells. Nature 518:207–212. CrossRef Medline grid cells is not sustained during reduced theta oscillations. Science 332: Sun C, Kitamura T, Yamamoto J, Martin J, Pignatelli M, Kitch LJ, Schnitzer 592–595. CrossRef Medline MJ, Tonegawa S (2015) Distinct speed dependence of entorhinal island Ko¨hler C (1985) Intrinsic projections of the retrohippocampal region in the and ocean cells, including respective grid cells. Proc Natl Acad Sci U S A rat brain: I. The subicular complex. J Comp Neurol 236:504–522. 112:9466–9471. CrossRef Medline CrossRef Medline Tang Q, Brecht M, Burgalossi A (2014a) Juxtacellular recording and mor- Kruge IU, Waaga T, Wernle T, Moser EI, Moser MB (2014) Grid cells re- phological identification of single neurons in freely moving rats. Nat Pro- quire experience with local boundaries during development: 2014 Neu- toc 9:2369–2381. CrossRef Medline roscience Meeting Planner (online 94.01/SS21). Washington, DC: Society Tang Q, Burgalossi A, Ebbesen CL, Ray S, Naumann R, Schmidt H, Spicher D, for Neuroscience. Brecht M (2014b) Pyramidal and stellate cell specificity of grid and bor- Krupic J, Burgess N, O’Keefe J (2012) Neural representations of location der representations in layer 2 of medial entorhinal cortex. Neuron 84: composed of spatially periodic bands. Science 337:853–857. CrossRef 1191–1197. CrossRef Medline Medline Tang Q, Ebbesen CL, Sanguinetti-Scheck JI, Preston-Ferrer P, Gundlfinger A, Krupic J, Bauza M, Burton S, Barry C, O’Keefe J (2015) Grid cell symmetry Winterer J, Beed P, Ray S, Naumann R, Schmitz D, Brecht M, Burgalossi is shaped by environmental geometry. Nature 518:232–235. CrossRef A (2015) Anatomical organization and spatiotemporal firing patterns Medline of layer 3 neurons in the rat medial entorhinal cortex. J Neurosci 35: Lever C, Burton S, Jeewajee A, O’Keefe J, Burgess N (2009) Boundary vector 12346–12354. CrossRef Medline cells in the subiculum of the hippocampal formation. J Neurosci 29: Taube JS (1995) Place cells recorded in the parasubiculum of freely moving 9771–9777. CrossRef Medline rats. Hippocampus 5:569–583. CrossRef Medline Liu P, Jarrard LE, Bilkey DK (2004) Excitotoxic lesions of the pre- and para- Taube JS (2007) The head-direction signal: origins and sensory-motor inte- subiculum disrupt the place fields of hippocampal pyramidal cells. Hip- gration. Annu Rev Neurosci 30:181–207. CrossRef Medline pocampus 14:107–116. CrossRef Medline van Groen T, Wyss JM (1990) The connections of presubiculum and para- Mitchell SJ, Rawlins JN, Steward O, Olton DS (1982) Medial septal area subiculum in the rat. Brain Res 518:227–243. CrossRef Medline lesions disrupt theta rhythm and cholinergic staining in medial entorhinal van Groen T, Wyss JM (1992) Projections from the laterodorsal nucleus of cortex and produce impaired radial arm maze behavior in rats. J Neurosci the thalamus to the limbic and visual cortices in the rat. J Comp Neurol 2:292–302. Medline 324:427–448. CrossRef Medline Mizuseki K, Sirota A, Pastalkova E, Buzsa´ki G (2009) Theta oscillations pro- van Strien NM, Cappaert NL, Witter MP (2009) The anatomy of memory: vide temporal windows for local circuit computation in the entorhinal- an interactive overview of the parahippocampal-hippocampal network. hippocampal loop. Neuron 64:267–280. CrossRef Medline Nat Rev Neurosci 10:272–282. CrossRef Medline Moser EI, Moser MB (2013) Grid cells and neural coding in high-end cor- Varga C, Lee SY, Soltesz I (2010) Target-selective GABAergic control of tices. Neuron 80:765–774. CrossRef Medline entorhinal cortex output. Nat Neurosci 13:822–824. CrossRef Medline Moser EI, Kropff E, Moser MB (2008) Place cells, grid cells, and the brain’s von Heimendahl M, Rao RP, Brecht M (2012) Weak and nondiscriminative spatial representation system. Annu Rev Neurosci 31:69–89. CrossRef responses to conspecifics in the rat hippocampus. J Neurosci 32: Medline 2129–2141. CrossRef Medline Mulders WH, West MJ, Slomianka L (1997) Neuron numbers in the pre- Wills TJ, Cacucci F, Burgess N, O’Keefe J (2010) Development of the hip- subiculum, parasubiculum, and entorhinal area of the rat. J Comp Neurol pocampal cognitive map in preweanling rats. Science 328:1573–1576. 385:83–94. CrossRef Medline CrossRef Medline Pastoll H, Solanka L, van Rossum MC, Nolan MF (2013) Feedback inhibi- Winter SS, Clark BJ, Taube JS (2015) Disruption of the head-direction cell tion enables theta-nested gamma oscillations and grid firing fields. Neu- network impairs the parahippocampal grid cell signal. Science 347: ron 77:141–154. CrossRef Medline 870–874. CrossRef Medline Paxinos G, Franklin KBJ (2012) Paxinos and Franklin’s the mouse brain in Zhang K, Sejnowski TJ (1999) Neuronal tuning: to sharpen or broaden? stereotaxic coordinates. San Diego: Academic. Neural Comput 11:75–84. CrossRef Medline Pinault D (1996) A novel single-cell staining procedure performed in vivo Zhang K, Ginzburg I, McNaughton BL, Sejnowski TJ (1998) Interpreting under electrophysiological control: morpho-functional features of juxta- neuronal population activity by reconstruction: unified framework with cellularly labeled thalamic cells and other central neurons with biocytin or application to hippocampal place cells. J Neurophysiol 79:1017–1044. Neurobiotin. J Neurosci Methods 65:113–136. CrossRef Medline Medline

65 Ebbesen (2017) Chapter 6 6. Cell Type-Specific Differences in Spike Timing and Spike Shape in the Rat Parasubiculum and Superficial Medial Entorhinal Cortex

This manuscript was published as: Ebbesen, C.L., Reifenstein, E.T., Tang, Q., Burgalossi, A., Ray, S., Schreiber, S., Kempter, R. & Brecht, M. (2016) Cell type-specific differences in spike timing and spike shape in rat parasubiculum and superficial medial entorhinal cortex.Cell Reports 16(4):1005-1015. Reprinted with permission from Elsevier under CC BY-NC-ND

66 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Article

Cell Type-Speciﬁc Differences in Spike Timing and Spike Shape in the Rat Parasubiculum and Superﬁcial Medial Entorhinal Cortex

Graphical Abstract Authors Christian Laut Ebbesen, Eric Torsten Reifenstein, Qiusong Tang, ..., Susanne Schreiber, Richard Kempter, Michael Brecht

Correspondence [email protected]

In Brief Neurons in the parahippocampal cortex discharge in elaborate spatiotemporal ﬁring patterns. Ebbesen et al. use juxtacellular recordings to show that the neuronal cell type is a major determinant of temporal discharge patterns such as bursting and phase precession.

Highlights

d We ﬁnd cell type-speciﬁc differences in spike shape, burstiness, and phase precession

d In vivo cell type speciﬁcity does not match predictions from previous in vitro studies

d Anatomical identity is a major determinant of spike patterns in the parahippocampal cortex

Ebbesen et al., 2016, Cell Reports 16, 1005–1015 July 26, 2016 ª 2016 The Authors. http://dx.doi.org/10.1016/j.celrep.2016.06.057

67 Ebbesen (2017) Chapter 6

Cell Reports Article

Cell Type-Speciﬁc Differences in Spike Timing and Spike Shape in the Rat Parasubiculum and Superﬁcial Medial Entorhinal Cortex

Christian Laut Ebbesen,1,2 Eric Torsten Reifenstein,1,3 Qiusong Tang,1 Andrea Burgalossi,4 Saikat Ray,1 Susanne Schreiber,1,3 Richard Kempter,1,3 and Michael Brecht1,* 1Bernstein Center for Computational Neuroscience, Humboldt Universita¨ t zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany 2Berlin School of Mind and Brain, Humboldt Universita¨ t zu Berlin, 10115 Berlin, Germany 3Institute for Theoretical Biology and Department of Biology, Humboldt Universita¨ t zu Berlin, 10115 Berlin, Germany 4Werner Reichardt Centre for Integrative Neuroscience, Otfried-Muller-Strass€ 25, 72076 Tubingen,€ Germany *Correspondence: [email protected] http://dx.doi.org/10.1016/j.celrep.2016.06.057

SUMMARY spatio-temporal firing patterns of defined cell types is needed to constrain models and help prune the forest of different The medial entorhinal cortex (MEC) and the adjacent models. Two aspects of the temporal firing patterns were high- parasubiculum are known for their elaborate spatial lighted in recent work: burstiness and theta cycle skipping. discharges (grid cells, border cells, etc.) and the pre- Burstiness has been shown to be associated with grid cell firing cessing of spikes relative to the local field potential. (Newman and Hasselmo, 2014; Latuske et al., 2015) and might We know little, however, about how spatio-temporal serve important functions in parahippocampal microcircuits firing patterns map onto cell types. We find that (Welday et al., 2011; Sheffield and Dombeck, 2015). Burstiness has also been linked to differences in extracellular spike shape cell type is a major determinant of spatio-temporal (Newman and Hasselmo, 2014; Latuske et al., 2015). Theta cycle discharge properties. Parasubicular neurons and skipping might be related to the computation of head-directional MEC layer 2 (L2) pyramids have shorter spikes, information and grid firing (Brandon et al., 2013). discharge spikes in bursts, and are theta-modulated Previous investigations of burstiness and theta cycle skipping (rhythmic, locking, skipping), but spikes phase-pre- have analyzed mixed extracellular recordings from both the cess only weakly. MEC L2 stellates and layer 3 (L3) superficial medial entorhinal cortex and the parasubiculum neurons have longer spikes, do not discharge in (Brandon et al., 2013; Newman and Hasselmo, 2014; Latuske bursts, and are weakly theta-modulated (non-rhyth- et al., 2015). It has thus remained unclear whether burstiness mic, weakly locking, rarely skipping), but spikes and theta cycle skipping map onto anatomical categories or steeply phase-precess. The similarities between whether bursty and non-bursty neurons are simply intermingled MEC L3 neurons and MEC L2 stellates on one hand (Latuske et al., 2015). Stellate cells (Stel) in layer 2 (L2) of the medial entorhinal cortex show a tendency to fire bursts of action and parasubicular neurons and MEC L2 pyramids potentials upon membrane depolarization in vitro (Alonso and on the other hand suggest two distinct streams of Klink, 1993; Pastoll et al., 2012; Alessi et al., 2016; Fuchs et al., temporal coding in the parahippocampal cortex. 2016). Such findings led to the hypothesis that stellate cells might display bursty firing patterns in vivo (Newman and Has- INTRODUCTION selmo, 2014; Latuske et al., 2015). Entorhinal grid cells phase-precess; i.e., they shift spike The discovery of grid cells in the medial entorhinal cortex (MEC) timing in a systematic way relative to the field potential during (Hafting et al., 2005) has been a major advance in cortical physi- firing field transversals (Hafting et al., 2008; Jeewajee et al., ology (Burgess 2014). The assessment of single-unit activity in 2013; Newman and Hasselmo 2014). Based on a pooled run rats running in boxes has led to the discovery of a plethora of analysis, it has been found that MEC L2 cells phase-precess ‘‘functional’’ cell types in the MEC: conjunctive (head-directional) more strongly than MEC layer 3 (L3) cells (Hafting et al., 2008; grid cells (Sargolini et al., 2006), border cells (Solstad et al., 2008), Mizuseki et al., 2009). This difference between MEC layers 2 boundary vector cells (Koenig et al., 2011), speed cells (Kropff and 3 has not been seen at the single run level; however, it et al., 2015), and cue cells (Kinkhabwala et al., 2015, J Neurosci., may arise because MEC L3 cells are less correlated between conference). Grid and border cells also exist in areas neighboring runs (Reifenstein et al., 2012, 2014). Recently, a single run the entorhinal cortex, such as the subiculum and pre- and parasu- analysis of phase precession revealed differences between biculum (Lever et al., 2009; Boccara et al., 2010; Tang et al., 2016). pyramidal and stellate neurons in MEC L2 (Reifenstein et al., Computational models propose many different mechanisms 2016). Parasubicular neurons provide specific input to MEC L2 to explain how grid cell discharges come about (Giocomo pyramidal neurons (Pyr) (Tang et al., 2016), but it is unknown et al., 2011; Zilli 2012). A better knowledge of the anatomy and whether parasubicular neurons phase-precess.

Cell Reports 16, 1005–1015, July 26, 2016 ª 2016 The Authors. 1005 This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

68 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Figure 1. Parasubicular and Superﬁcial Medial Entorhinal Cortex Neuron Types (A) Top: tangential section of the parasubiculum (PaS) and layer 2 of the medial entorhinal cortex (MEC) stained for calbindin (Cb, green channel) and wolframin (WFS1, red channel). Bottom: parasagittal section of the MEC stained for Cb (green channel) and PCP4 (red channel). Also visible are the presubiculum (PrS) and postrhinal cortex (Por). (B) Left: reconstructions (from tangential cortical sections; neurons are seen from the top) of examples of the four neuron types: a PaS neuron (blue), an MEC L2 pyramidal neuron (green), an MEC L2 stellate cell (black), and an MEC L3 neuron (red), corresponding to the anatomical cell types marked by arrows in (A). Right: juxtacellular recording traces of the reconstructed cells. The spiking of the parasubicular neuron and the MEC L2 pyramid is bursty and theta-modulated. Scale bars, 1 mV. Cell reconstructions were adapted from Tang et al. (2014a, 2015, 2016).

Here we analyze juxtacellular recordings from the medial en- specifically target the patches of MEC L2 pyramidal cells (Burga- torhinal cortex (Ray et al., 2014; Tang et al., 2014a, 2015) and lossi et al., 2011; Tang et al., 2016). MEC L3 neurons are not ar- the parasubiculum (Tang et al., 2016). Juxtacellular data offer ranged in a hexagonal grid but are visible as a homogenous band two advantages (Pinault 1996; Herfst et al., 2012). First, of Purkinje cell protein 4 (PCP4)-positive cells below layer 2 (L3; cells can often be anatomically identified. Second, juxtacellular red band in Figure 1A, bottom). Figure 1B, left, shows recon- recording of the local field potential (LFP) and spikes has a very structions of example cells of the four neuron types: a parasubic- high temporal resolution and signal-to-noise ratio, which is ular neuron (blue), a MEC L2 pyramidal neuron (green), a MEC L2 crucial for investigating temporal patterns such as burstiness. stellate cell (black), and an MEC L3 pyramidal neuron (red), all re- We ask the following questions. Does burstiness differ between corded in freely moving rats. We use these colors throughout the parasubicular neurons, MEC L2 pyramids, MEC L2 stellates, manuscript. All reconstructions are from tangential sections (i.e., and MEC L3 neurons? Are MEC L2 stellates actually bursty a ‘‘top view’’ of the morphology). In addition to the morphology, in vivo? Do differences in extracellular spike shape reflect bursti- we also show juxtacellular recording traces from the recon- ness or anatomical category? Does theta cycle skipping map structed example cells (Figure 1B, right). Two signals are visible onto anatomical categories? Does burstiness predict theta in the recordings: the spikes of the identified cells and the prom- rhythmicity and theta locking? How does phase precession inent theta rhythm in the LFP. differ among cell types? Analysis of Burstiness RESULTS To determine whether a neuron was discharging in a bursty pattern, we analyzed the interspike interval (ISI) histogram using Overview of Anatomical Cell Types in the a similar approach as Latuske et al. (2015). ISIs below 60 ms Parahippocampal Cortex were binned in 2-ms bins (normalized to area = 1 to generate a The parahippocampal cortex has a modular architecture. L2 of probability distribution), which revealed that our dataset con- the MEC contains patches of calbindin-positive pyramidal neu- tained both non-bursty and bursty cells (Figure 2A). We per- rons arranged in a hexagonal grid (Ray et al., 2014; Figure 1A, formed a principal component analysis on a matrix of the ISI top) that are surrounded by calbindin-negative stellate cells (Fig- probability distributions of all neurons and found that the first ure 1A, top, black background). The parasubiculum (PaS) is a three principal components (PCs; Figure 2B, bottom) explained thin elongated structure that wraps around the MEC mediodor- 69% of the variance in the data. In agreement with Latuske sally and has high wolframin expression (WFS1-positive cells; et al. (2015), we found that, when the first two principal compo- Figure 1A, top; Tang et al., 2016). Axons from the parasubiculum nents were plotted against each other, the neurons formed a

1006 Cell Reports 16, 1005–1015, July 26, 2016

69 Ebbesen (2017) Chapter 6

Figure 2. Classification of Bursty and Non-bursty neurons (A) Example ISI distribution of a bursty (left) and non-bursty (right) juxtacellularly recorded neuron (bin width, 2 ms). (B) Top: scatterplot of the first two principal components (PC1 and PC2) obtained from a PCA of ISI distributions (black dots). The neurons form a C-shaped structure, as described by Latuske et al. (2015) (2D kernel smoothed density estimate indicated by lines). Bottom: the first three PCs of the ISI histograms. (C) Top: 3D scatterplot of the first three PCs, assigned to two clusters using a k-means clustering algorithm. Center-of-mass of bursty neurons (orange) and non- bursty neurons (purple) are indicated by black crosses. Bottom: projection of ISI distributions onto the optimal linear discriminant (the burstiness) of the two clusters revealed a bimodal distribution of bursty (orange) and non-bursty (purple) neurons. (D) Left: ISI histograms of all classified neurons, sorted by burstiness (scaled to maximum probability for each neuron for visibility). Right: example ISI histograms of neurons at the edges and in the middle of the clusters. Bursty neurons tend to fire burst at 125–250 Hz (4- to 8-ms intervals).

C-shaped structure, indicative of a bimodal distribution (Fig- MEC L3 neurons were flat with no obvious burstiness (Figure 3A, ure 2B, top). bottom). To assess whether this difference was statistically signif- We assigned the neurons to two clusters using a k-means icant, we performed two tests: one based on categorical classifi- clustering algorithm on the first three principal components cations of cells as ‘‘non-bursty’’ and ‘‘bursty’’ with a guard zone (Figure 2C, top). The two clusters were well separated with little (Experimental Procedures; Latuske et al., 2015) and another one overlap (Figure 2D). To assess the separation quality of the two where we directly compared burstiness among the neuron types. clusters, we calculated the projection of the neurons onto When we compared the proportions of non-bursty, guard- Fisher’s linear discriminant. We can interpret the linear discrimi- zoned, and bursty cells among neuron types, we found no signif- nant as a measure of ‘‘burstiness’’ because it is places the cells icant difference between parasubicular neurons and MEC L2 along an axis from non-bursty to bursty based on the shape of pyramids, which both contained predominantly bursty cells the ISI histogram. We plotted all cells sorted according to bursti- (PaS versus Pyr, bursty/guard/non-bursty: 11/11/0 versus 15/ ness, and, in agreement with Latuske et al. (2015), we found that 15/1, p > 0.05, c2 test; Figure 3B). We also found no difference bursty neurons were distinguished by a tendency to fire bursts at between MEC L2 stellate cells and MEC L3 cells (Stel versus 125–250 Hz (4- to 8-ms bins; Figure 2D). L3, bursty/guard/non-bursty: 9/25/34 versus 3/5/24, p > 0.05, To investigate differences in burstiness among cell types, we c2 test; Figure 3B), which were both predominantly non-bursty. plotted the median ISI histogram of all recorded cells, resolved Both parasubicular neurons and MEC L2 pyramids contained by cell type. The median ISI histograms of parasubicular as well significantly different proportions of bursty and non-bursty cells as MEC L2 pyramidal neurons indicated very bursty cells (Fig- in comparison with both MEC L2 stellates and MEC L3 neurons ure 3A, top). The median ISI histograms of MEC L2 stellate and (all p < 0.001, c2 tests; Figure 3B).

Cell Reports 16, 1005–1015, July 26, 2016 1007

70 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Figure 3. Burstiness in the Parasubiculum and Superficial Medial Entorhinal Cortex (A) Median ISI histogram (bin width, 2 ms) of all neurons recorded in the PaS (blue), identified and putative MEC L2 pyramidal neurons (green), identified and putative MEC L2 stellate cells (black), and MEC L3 neurons (red). Grey lines indicate 95% confidence intervals of the median. (B) Comparison of the proportions of the numbers of bursty (orange) and non-bursty (purple) neurons for the four different neuron types defined in (A). White areas denote cells that fall in the ambiguous zone between non-bursty and bursty (c2 tests of equal proportions among cell types). (C) Comparison of the burstiness for the four different neuron types defined in (A). Vertical lines indicate medians (Mann-Whitney U tests).

Using a categorical classifier with a guard zone has potential result of the classification method. First we used a statistical problems. The width and placement of the guard zone is esti- method. We fitted three generalized linear models to investigate mated from the bimodal fit, and thus the guard zone depends whether burstiness might be related to theta strength (model 1, on the relative abundance of bursty and non-bursty cells, which burstiness strength; Figure S3A, left), putative cell type is evidently not the same among neuron types; i.e., the guard (model 2, burstiness type; Figure S3A, middle), or both (model 3, zone might be either too wide or too narrow. The guard zone burstiness type + strength; Figure S3A, right). Both compari- also discards information telling us whether a neuron is near sons of the Akaike information criterion (AIC; Akaike, 1974) and the guard zone or closer to the extremes. These problems may likelihood ratio tests of nested models indicated that model 2 inflate our estimated differences in burstiness among cell types. is superior to the other models (Figure S3B); i.e., the bursti-

To make sure that no spurious results were imposed by the ness depends only on putative cell type (model 2, PType = guard zone, we directly compared the burstiness of the neuron 0.0000076; Figure S3C, middle) and not on theta strength types and included all cells. In agreement with the estimations (model 2 versus model 3, p = 0.54, likelihood ratio test; Fig- based on comparisons of the proportions, we found that the ure S3B). Second, we plotted the burstiness among cell types burstiness of parasubicular neurons and MEC L2 pyramids twice: once where we include the classified MEC L2 cells (Fig- was significantly higher than the burstiness in both MEC L2 stel- ure S3D, left) and once where we only include identified MEC lates and MEC L3 neurons (all p < 0.001, Mann-Whitney U tests; L2 cells (Figure S3D, right). The pattern of burstiness among Figure 3C). Again, we did not find a significant difference be- cell types remained the same when we only included the identi- tween parasubicular neurons and MEC L2 pyramids (p > 0.05, fied cells (Figure S3D). We thus conclude that cell type-specific Mann-Whitney U test; Figure 3C), but we did find that MEC L3 differences in burstiness are not an artifact of our classification neurons had a significantly lower burstiness than MEC L2 stel- approach. lates (p = 0.0036, Mann-Whitney U test; Figure 3C). Thus, parasubicular neurons and MEC L2 pyramids are gener- Analysis of Spike Shape ally bursty, whereas MEC L2 stellates and MEC L3 neurons are In tetrode recordings of parasubicular and MEC L2/3 neurons, generally non-bursty (Figures 3A and 3B). Furthermore, within differences in spike shape have been linked to burstiness (La- the non-bursty neuron types, MEC L3 neurons are more strictly tuske et al., 2015) and theta phase preference of grid cells (New- non-bursty than MEC L2 stellates (Figure 3C). It should be noted, man and Hasselmo, 2014). We therefore investigated whether however, that even though there are large and highly significant there was a difference in spike shape among our four anatomical differences in burstiness among cell types, the distributions of categories of neurons. First we removed a subset of cells for burstiness among cell types are overlapping. For example, a mi- which the signal-to-noise ratio of spike waveforms was insuffi- nority of L2 stellate cells and L3 neurons assume firing patterns cient to reliably assess the spike shape. Second, we removed that are otherwise classically parasubicular/pyramid-like. spikes that happened within 100 ms of the previous spike to Our dataset includes MEC L2 neurons that were classified as disregard potential effects of spike shape adaptation during putatively pyramidal or stellate based on theta strength and bursts (Experimental Procedures). In Figure 4A, we plot the preferred theta phase (Tang et al., 2014a; Figure S1). We there- remaining spike shapes (normalized for display; Experimental fore also checked whether there was any correlation between Procedures) for all four neuron types. We did not find any differ- burstiness and theta strength because such a correlation might ences among neuron types in spike amplitude, peak-to-trough introduce ‘‘artifactual’’ cell type differences in burstiness as a ratio, or spike half-width (all p > 0.05, Kruskal-Wallis test). This

1008 Cell Reports 16, 1005–1015, July 26, 2016

71 Ebbesen (2017) Chapter 6

Figure 4. Spike Shapes in the Parasubicu- lum and Superficial Medial Entorhinal Cortex (A) Peak-aligned and voltage-scaled spike shapes of cells in the PaS (blue), identified and putative MEC L2 pyramidal neurons (green), identified and putative MEC L2 stellate cells (black), and MEC L3 neurons (red). (B) Left: mean spike shapes of the four neuron types in (A) show differences in peak-to-trough time. Right: close-up of the trough of the mean spike shapes. (C) Comparison of peak-to-trough times of neurons as defined in (A) (Mann-Whitney U test; horizontal lines indicate means).

as a function of only neuron type (GLM2, peak-to-trough 1 + type), we also found a signiﬁcant dependence on neuron type

(ANOVA, pType = 0.0015; Figure S4A, solid lines). However, when we modeled peak-to-trough time as a function of was expected for two reasons: Overall spike amplitude depends both burstiness and neuron type (GLM3, peak-to-trough 1+ strongly on the particular recording pipette and relation to burstiness + type), we found that the dependency on type but the soma (Gold et al., 2009), and narrow spikes and a small not the dependency on burstiness remained signiﬁcant (ANOVA,

peak-to-trough ratio are indicative of interneurons (Mountcastle pBurstiness = 0.22, pType = 0.017; Figure S4C). We also fitted a et al., 1969; Csicsvari et al., 1999), and we consider here four model where we allowed for interactions between burstiness types of excitatory principal cells. and type (GLM4, peak-to-trough 1 + burstiness + type + bursti- We noticed, however, a large variability in the repolarization ness*type), where all effects became non-significant (ANOVA, all phase of the cell type: Parasubicular neurons and MEC L2 pyra- p > 0.05; Figure S4C). To determine which model best explains mids contained many cells that quickly reached the trough and the data, we calculated the AIC of all models and found that, repolarized, whereas MEC L2 stellates and MEC L3 neurons despite the four fitted parameters, GLM2 had the lowest AIC, reached the trough more slowly (Figure 4A). This tendency was indicating that the peak-to-trough time depends on neuron also evident in the mean spike shape of the four neuron types type, but not on burstiness (Figure S4B). Similarly, when (Figure 4B). When we compared the peak-to-trough time of the comparing nested models, we found that GLM3 better explains cell types, we found significant differences (p = 0.0014, Krus- the data than GLM1 (p = 0.0023, likelihood ratio test; Figure S4B); kal-Wallis test). Parasubicular neurons and MEC L2 pyramids i.e., including neuron type as a predictor makes the model better. had significantly shorter peak-to-trough times than both MEC We did not find that GLM3 explains the data better than GLM2 L2 stellates and MEC L3 neurons (all p < 0.05, Mann-Whitney (p = 0.32, likelihood ratio test; Figure S4B); i.e., it is unnecessary U test; Figure 4C). to include burstiness as a predictor in addition to neuron type. We thus infer that the differences in spike shape primarily reflect Is Spike Shape a Reflection of Burstiness or Cell Type? the anatomical type and not the burstiness of the neuron. Because parasubicular neurons and MEC L2 pyramids have faster peak-to-trough times and are also the most bursty cell Analysis of Rhythmicity and Theta Cycle Skipping types, we wondered whether, as has been suggested (Latuske To determine whether a neuron was theta cycle-skipping, we et al., 2015), the burstiness of the cell predicts the spike shape. used a maximum likelihood estimation (MLE) of a parametric Alternatively, the spike shape might simply be different among model of the ISI histogram (Climer et al., 2015; Experimental Pro- neuron types, or it might depend on neuron types as well as cedures). Our dataset contained neurons that showed no theta burstiness. To figure this out, we decided to employ a general- modulation and also neurons that had strong rhythmic compo- ized linear regression approach. Because peak-to-trough time nents (Figure 5A). For every cell, we fitted three models to the cannot assume negative values, we modeled peak-to-trough ISI distribution: a ‘‘flat’’ model with no rhythmic components (Fig- time as a gamma-distributed variable (Experimental Proce- ure 5A, left), a ‘‘rhythmic, non-skipping’’ model with a theta- dures). We selected the appropriate model using the following rhythmic modulation of the ISI histogram (Figure 5A, middle), approach: We first modeled peak-to-trough time as a function and a ‘‘rhythmic, cycle-skipping’’ model with a theta-rhythmic of only burstiness (GLM1, peak-to-trough 1 + burstiness) and modulation of the ISI histogram and a second parameter intro- found a significant dependence (ANOVA, pBurstiness = 0.0087; ducing theta cycle skipping (i.e., a higher amplitude of every Figure S4A, dashed gray line). This result is in agreement with La- other peak in the ISI histogram; Figure 5A, right). The three tuske et al. (2015). Also, when we modeled peak-to-trough time fitted models were compared using the appropriate c2 statistic

Cell Reports 16, 1005–1015, July 26, 2016 1009

72 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Figure 5. Theta Rhythmicity and Theta Cycle Skipping in the Parasubiculum and Superficial Medial Entorhinal Cortex (A) Example ISI histograms (black bars) of non- rhythmic (left), rhythmic and non-skipping (middle), and rhythmic but theta cycle-skipping (right) juxtacellularly recorded neurons. Solid red lines show maximum likelihood estimates of the ISI, and dashed blue lines indicate a flat model (no rhythmicity or cycle skipping). Bin width, 1 ms. (B) Flow diagram of the cell classification procedure. First we checked for rhythmicity and then for cycle skipping. (C) Left: comparison of the proportions of non- rhythmic and rhythmic neurons recorded in the PaS, identified and putative MEC L2 pyramidal neurons, identified and putative MEC L2 stellate cells, and MEC L3 neurons. Right: comparison of the proportions of rhythmic, non-cycle-skipping and rhythmic, theta cycle-skipping neurons recorded in the four neuron types. The generally rhythmic cell types (PaS and Pyr) have a larger proportion of theta cycle-skipping neurons than the generally non-rhythmic cell types (Stel and L3).

(calculated from the maximum log likelihood of the models) to Climer et al. (2015) returns a p value of the rhythmicity per cell

generate two p values: prhythmic (comparing the flat and the rhyth- and is sensitive to very low amounts of rhythmicity, which could mic, non-skipping models) and pskipping (comparing the rhyth- potentially have been present in, e.g., putative stellates with a mic, non-skipping and the rhythmic, cycle-skipping models). low locking strength and locking to the peak of the LFP theta The cells were classified using a two-level classification (Fig- rhythm (Figure S1; Climer et al., 2015; Tang et al., 2014a). ure 5B): First, we determined whether a cell was ‘‘rhythmic’’ More importantly, our classification procedure considers simply

(prhythmic < 0.05) or ‘‘non-rhythmic’’ (prhythmic > 0.05). Then we strength of locking to the local LFP, and there is no way of distin- classiﬁed the rhythmic cells as either rhythmic, cycle-skipping guishing a simply theta-rhythmic cell from a rhythmic and cycle-

(pskipping < 0.05) or rhythmic, non-skipping (pskipping > 0.05). skipping cell based on theta strength because they might show Using the MLE approach, we found that parasubicular neu- equally strong locking. To be sure that the cell type differences rons and MEC L2 pyramids were overwhelmingly rhythmic were not an artifact of including the classified cells, we plotted ( 93%; PaS, 20/22; Pyr, 29/31; Figure 5C, left). MEC L2 stellates the burstiness among cell types twice: once where we included and MEC L3 neurons were rarely rhythmic ( 26%; Stel, 16/68; the classified MEC L2 cells (Figure S5A, left) and once where we L3, 9/32), both significantly less rhythmic than both parasubicu- only included identified MEC L2 cells (Figure S5A, right). The pro- lar neurons and MEC L2 pyramids (all p < 0.001, c2 tests; Fig- portions among cell types remained the same when restricting ure 5C, left). This is in agreement with previous observations in the analysis to identified cells only (Figure S5A). which evaluated spike train rhythmicity of cell types using a ’’theta index’’ was used (Ray et al., 2014; Tang et al., 2014a, Single Run Analysis of Phase Precession 2016). We found that the generally rhythmic cell types were To compare the magnitude of phase precession among cell also significantly more likely to also be theta cycle-skipping types at the single-run level, we first selected single runs of than the generally non-rhythmic cell types (p = 0.018, mixed-ef- high firing based on the firing rate (Figures 6A, top, and 6B; fects logistic regression; Figure 5C, right; Experimental Proce- Experimental Procedures). From these single runs, we deter- dures): Approximately 49% of the rhythmic parasubicular neu- mined the slope and range of phase precession by a circular- rons and rhythmic MEC L2 pyramids were also theta cycle- linear fit of time and theta phase angle of the spikes in each skipping (PaS, 9/20; Pyr, 15/29; Figure 5C, right). Of the MEC run (Figure 6A, bottom; Experimental Procedures). Figure 6C L2 stellates and MEC L3 neurons, which were classified as rhyth- shows example single runs from example cells of the four neuron mic using the MLE approach, only 20% were also theta cycle- types. The example MEC L2 stellate and L3 neurons have steep skipping (Stel, 4/16; L3, 1/9; Figure 5C, right). phase precession slopes and cover larger ranges of theta phase Our dataset includes MEC L2 neurons that were classified angles during a single run. In contrast, the example parasubicu- as putatively pyramidal or stellate based on theta strength and lar neuron and MEC L2 pyramid only weakly phase-precess. preferred theta phase (Tang et al., 2014a; Figure S1). Obviously, Across the population, we found the same result: First, identified we expect a correlation between the theta rhythmicity (which is and putative MEC L2 stellate and L3 neurons had approximately calculated from the ISI distribution) and the theta strength (lock- 3-fold steeper phase precession slopes than parasubicular neu- ing to the LFP theta rhythm). However, the MLE approach of rons and identified and putative MEC L2 pyramids (Figure 6D;

1010 Cell Reports 16, 1005–1015, July 26, 2016

73 Ebbesen (2017) Chapter 6

Figure 6. Phase Precession Slopes and Ranges in the Parasubiculum and Superficial Medial Entorhinal Cortex (A) Detection of single runs. Top: firing rate (red line) is estimated by convolving spikes (blue ticks) with a Gaussian kernel. Detected runs are indicated by gray shading. Bottom: theta phase of spikes as a function of time (black dots). Phase precession slopes and ranges of single runs are estimated by circular-linear fits (dashed lines). (B) Temporally defined single runs (black lines) match regions of elevated firing rate (color coded). Data are from the neuron shown in (A). (C) Examples of single-run phase precession for parasubicular (blue dots), identified MEC L2 pyramidal (green dots), identified MEC L2 stellate (black dots), and MEC L3 (red dots) neurons. Each dot represents the theta phase angle of a spike as a function of time. Dashed lines depict circular-linear fits. (D) Median single-run phase precession slopes for the four neuron types defined in (C). Single-run slopes are significantly larger in MEC L2 stellate and MEC L3 neurons than in parasubicular and MEC L2 pyramidal neurons (whiskers indicate 95% confidence intervals of the median). (E) Median single-run phase precession ranges among the four neuron types as defined in (C) and (D). Single-run phase ranges are significantly larger in MEC L2 stellate and MEC L3 neurons than in parasubicular and MEC L2 pyramidal neurons (whiskers indicate 95% confidence intervals of the median).

median slopes: PaS/Pyr/Stel/L3 = 16.7/ 25.9/ 76.7/ 64.8 entorhinal stellate neurons, and parasubicular regular spiking À À À À degrees/s; p(PaS versus Stel) = 0.0001; p(PaS versus L3) = cells. 0.003; p(Pyr versus Stel) = 0.0001; p(Pyr versus L3) = 0.01; Mann-Whitney U tests). Second, identified and putative MEC Cell Type-Specific Differences and Their Origin L2 stellate and L3 neurons covered a much larger range of theta We found significant differences in spike shape, burstiness, phase angles per run than parasubicular neurons and identified theta modulation (rhythmicity, locking, cycle skipping, phase and putative MEC L2 pyramids (approximately 2-fold; Fig- precession), and theta phase precession between the four ure 6E; median ranges: PaS/Pyr/Stel/L3 = 63.2/48.7/127.5/ groups of cells investigated. Thus, our data suggest that cell 114.2 degrees; p(PaS versus Stel) = 0.00008; p(PaS versus type is a major determinant of discharge patterns in the rat para- L3) = 0.0007; p(Pyr versus Stel) = 0.0000002; p(Pyr versus subiculum and superficial medial entorhinal cortex. Although our L3) = 0.00005; Mann-Whitney U tests). We did not find any differ- data emphasize the significance of cell types, the discharge ences in the circular-linear correlation coefficient among the cell patterns we observed do not directly match what is expected types (p = 0.38, Kruskal-Wallis test). based on the analysis of intrinsic properties of these neurons in vitro. In vitro recordings of parasubicular neurons have sug- DISCUSSION gested an intrinsic disposition for theta rhythmicity (Glasgow and Chapman, 2008). It is known that in vitro measurements of We used advanced statistical techniques to tease apart L2 MEC cell properties are very sensitive to recording conditions how differences in burstiness, spike shape, theta modula- (Alonso and Klink, 1993; Pastoll et al., 2012). However, MEC L2 tion (rhythmicity, locking, skipping), and phase precession stellates often display some intrinsic burstiness in vitro (Alonso map onto regular spiking layer 3 medial entorhinal neurons, and Klink, 1993; Pastoll et al., 2012; Alessi et al., 2016; Fuchs layer 2 medial entorhinal pyramidal neurons, layer 2 medial et al., 2016), but they are generally not very bursty in vivo (Ray

Cell Reports 16, 1005–1015, July 26, 2016 1011

74 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

et al., 2014; Figure 2). Thus, it is probably incorrect to assume erties of the entorhinal network. From these studies we know that that bursty cells recorded extracellularly in the superficial MEC grid cells are bursty whereas border cells are not (Newman and and the parasubiculum are MEC L2 stellates (Newman and Has- Hasselmo 2014; Latuske et al., 2015). It has also been shown selmo, 2014; Latuske et al., 2015) because we show that bursty that theta cycle skipping is somehow necessary for maintaining cells are more likely to be MEC L2 pyramids or parasubicular grid cell firing (Brandon et al., 2013). In agreement with Tang et al. neurons. (2014a, 2016), we conclude that, based on burstiness and theta cycle skipping, parasubicular neurons and MEC L2 pyramids are Cell Type Specificity of Phase Precession likely to play a key role in generating grid cell activity in the para- Although phase precession is arguably the most intensely stud- subiculum and superficial medial entorhinal cortex. ied example of temporal coding in the brain, its underlying mechanism is still a matter of debate. Parasubicular neurons, which Cell Type-Specific Differences in Spike Shape show only weak phase precession, project to pyramidal cells in In line with the differences in temporal discharge patterns, we MEC L2 (Tang et al., 2016). Also, these MEC L2 pyramidal cells observed that parasubicular and MEC L2 pyramidal cells had express only a low degree of phase precession. Conversely, stel- shorter spike durations than MEC L2 stellates and MEC L3 neu- late cells in MEC L2 and pyramidal cells in MEC L3 phase pre- rons. Several previous studies have noticed significant differ- cess with steep slopes. The latter finding is somewhat surprising ences between MEC L2 pyramidal and MEC L2 stellate cells, because it challenges the long-held belief that cells in MEC L3 most notably, that stellate cells have larger depolarizing afterpo- do not phase-precess (Hafting et al., 2008; Mizuseki et al., tentials (Alonso and Klink 1993; Alessi et al., 2016; Fuchs et al., 2009). However, differences in methodology might reconcile 2016). In in vivo recordings, it was generally observed that stel- the different findings. Previous studies investigated MEC L3 late cells had a shorter spike duration than pyramidal cells phase precession in pooled run data. In contrast to that, we (Alonso and Klink, 1993). Interestingly, however, it was also analyzed phase precession in single runs (Schmidt et al., found that the spike duration of both pyramidal and stellate cells 2009). We argue that the single-run approach is more appro- varied depending on the depolarizing current pulse (Alonso and priate because the animal needs to process information online Klink, 1993). Thus, the juxtacellularly observed differences in and does not have the opportunity to pool over trials. Our finding spike shape are probably not primarily a reflection of differences of substantial MEC L3 phase precession is in line with a previous in intrinsic cell properties. Cell type differences in spike duration single-run account (Reifenstein et al., 2014). MEC L2 stellate are statistically significant. However, the distributions of spike cells project to the dentate gyrus, whereas MEC L2 pyramidal durations are largely overlapping (Figure 4C), probably preclud- cells send output to CA1 (Varga et al., 2010; Kitamura et al., ing a classification of extracellularly recorded MEC L2 regular 2014; Ray et al., 2014). Because MEC L2 pyramidal cells show spiking neurons into pyramidal and stellate cells based purely only weak phase precession, it seems unlikely that they substan- on spike shape. tially contribute to CA1 phase precession. Therefore, CA1 either generates phase precession de novo or inherits phase-precess- Do Layer 3 Cells and Layer 2 Stellate Cells, on One Hand, ing inputs via the strongly precessing stellate cells in MEC L3 and Parasubiculum and Layer 2 Pyramids, on the Other (Jaramillo et al., 2014). Hand, Form Two Distinct Processing Systems? We observed a strong similarity between spike shapes and firing Whether Cell Types Show Specific Spatial Discharge patterns of parasubicular neurons and MEC L2 pyramids. These Patterns Is Currently Unresolved two neuron groups were different in spike shapes and firing pat- It is presently unknown how the functional categories (grid cells, terns from layer 3 cells and layer 2 stellate cells, which were border cells, speed cells, cue cells, etc.) map onto the anatomy. similar to each other, however. It turns out that these neurons For example, it is unknown whether MEC L2 grid cells are pre- groups share even more similarities and differences. Parasubic- dominantly pyramidal cells (Tang et al., 2014a) or stellate cells ular axons specifically target patches of MEC L2 pyramidal cells (Domnisoru et al., 2013) or whether they show no preference (Tang et al., 2016), which might be a pathway for head-direc- for either cell type (Sun et al., 2015). Similarly, some authors tional information from the medial septum to reach the grid cell have reported that about a third to half of MEC L3 neurons are system (Winter et al., 2015; Unal et al., 2015; Tang et al., grid cells (Sargolini et al., 2006; Boccara et al., 2010), whereas 2016). L3 cells and layer 2 stellate cells provide a massive direct others have estimated that if L3 grid cells exists, then they (L3) and indirect input to the hippocampus, whereas projections must be rare ( 1%; Tang et al., 2015). from both layer 2 pyramids and the parasubiculum are minor or absent (Varga et al., 2010; Ray et al., 2014; Kitamura et al., 2014). Relation between Temporal Spiking Features and Thus, analysis of spike shapes and firing patterns, direct connec- Spatial Responses tivity, and projection targets supports the distinction of layer 3 Parasubicular neurons and MEC L2 pyramids are more bursty, cells and layer 2 stellate cells on one hand and parasubiculum have narrower spikes, and are more likely to be theta-rhythmic, and layer 2 pyramids on the other hand as two distinct process- theta-locked, and theta cycle-skipping than MEC L2 stellates ing systems. and MEC L3 neurons. These differences remain even when we statistically control for interactions between spike shape, bursti- Possible Anatomical Origin of Firing Patterns ness, and rhythmicity. Some studies have tried to elucidate the Layer 2 pyramids and parasubicular cells are anatomically grid cell generation mechanism by characterizing the firing prop- similar. They both express wolframin (Ray and Brecht, 2016),

1012 Cell Reports 16, 1005–1015, July 26, 2016

75 Ebbesen (2017) Chapter 6

and, in the early development stages, they also express calbin- refer to the pooled groups of identified and putative calbindin-positive pyrami- din (Ray and Brecht, 2016). Likewise, layer 3 neurons and layer dal cells simply as ‘‘MEC L2 pyramids’’ and identified putative calbindin-nega- 2 stellate cells also have an anatomical likeliness in their protein tive stellate cells as ‘‘MEC L2 stellates.’’ When we show example cells of the four cell types (Figures 1A and 1B and 6A–6C), we show only identified expression profile, with both expressing Reelin in adult rats (Ray Cb+/ cells. In Figures S3 and S5, we show analysis of a dataset where we À and Brecht, 2016). This might allude to the electrophysiological only included identified cells. and functional characteristics of these two groups being perhaps somewhat genetically determined, with the protein Analysis of Burstiness expression profiles of these respective cell groups shaping their To determine whether a neuron was discharging in a bursty pattern, we inputs and outputs. analyzed the ISI histogram using a similar approach as Latuske et al. (2015). ISIs below 60 ms were binned in 2-ms bins and normalized to area = 1 to generate a probability distribution (Figure 2A). A principle component analysis Grid Cell Models (PCA) was done on a matrix of the ISI probability distribution of all neurons Our results will constrain future modeling of network activity in (‘‘pca’’ in MATLAB, MathWorks). For plotting, the density of cells in this space the hippocampus and para-hippocampal cortices. Because was estimated with a 2D Gaussian kernel density estimator (‘‘kde2d’’; Botev different anatomical cell types have different projection patterns, et al., 2010). The neurons were assigned to two clusters using a k-means clus- burstiness, and theta rhythmicity/skipping might be passed on tering algorithm on the first three principal components (‘‘kmeans’’; MATLAB; differentially to hippocampal subfields like the dentate gyrus, Figure 2C, top). To assess the separation quality of the two clusters, we calculated the projection of the neurons onto Fisher’s linear discriminant (the bursti- which receives massive MEC L2 stellate input (Varga et al., ness, using ‘‘LDA’’ from Scikit-Learn in Python) and found that the two clusters 2010), and CA1, which receives some MEC L2 pyramidal input (non-bursty and bursty) were well separated with little overlap (Figure 2C, top). (Kitamura et al., 2014). Some grid cell models suggest that grid To check whether the distribution of burstiness was bimodal, thus reflecting cells are generated by network mechanisms where a large two distinct classes of ISI histograms, we fitted probability density functions number of similar (stellate) cells self-organize to generate sym- for Gaussian mixture models with between one and three underlying Gauss- metrical firing patterns either via continuous attractors or via ians and compared the models using the Akaike information criterion (Akaike 1974; AIC from ‘‘gmdistribution.fit’’ in MATLAB). A bimodal distribution best oscillatory interference (for reviews, see Giocomo et al., 2011; explained the data (AICunimodal = 622.7, AICbimodal = 609.6, AICtrimodal = Zilli 2012). Others have suggested mechanisms based on 614.7). Based on the mean and variance of the two Gaussian distributions un- anatomical microcircuits (Brecht et al., 2013). Our results do derlying the observed distribution of burstiness (Figure 2C, bottom, dashed not resolve this question, but we add to the picture that red lines), we estimated that excluding cells where 0.4 < burstiness < 1.5 À the network mechanism distributes firing patterns differentially would yield >95% correct labeling of non-bursty and bursty neurons in the according to cell type. non-bursty and bursty categories and used this as a guard zone (Latuske et al., 2015).

Conclusions Analysis of Spike Shape We conclude that the anatomical identity of the neuron is a During recording, the juxtacellular traces were digitized at 20 kHz. To analyze strong determinant of the firing pattern. Analysis of burstiness, the spike shapes, we first zero-phased high pass-filtered the raw signal at theta cycle skipping, and phase precession jointly suggest sim- 100 Hz with a finite impulse response filter of order 28 (‘‘fir1’’ in MATLAB). ilarities between layer 3 cells and layer 2 stellate cells on one The spike times were detected by thresholding the filtered signal and saving each threshold crossing ± 2.5 ms. Spike sorting based on the first principal hand and layer 2 pyramidal cells and parasubicular cells on the components was performed on these 5-ms snippets to remove any threshold other hand. crossings because of artifacts in the signal (Tang et al., 2014a). To align the spike shapes optimally after spike sorting, the 5-ms snippets were over- EXPERIMENTAL PROCEDURES sampled at five times their original sampling rate using a spline interpolation (‘‘interp1’’ in MATLAB) and were then aligned to the peak sample. To ensure All experimental procedures were performed according to the German guide- that we were only analyzing shapes free of distortions because of drift of the lines on animal welfare under the supervision of local ethics committees. pipette and that the spikes were well above the noise floor, we only analyzed spikes for which the spike amplitude was in the top 60th-90th percentile and Juxtacellular Recordings and Immunohistochemistry where the Z score of the spike amplitude was >17. The noise floor was defined In this paper, we analyzed a dataset of juxtacellular recordings from the super- as the mean of the first and last 0.5 ms of each 5-ms spike snippet. We ficial medial entorhinal cortex and the parasubiculum that we have published also removed any spikes where there was another spike in the preceding previously (Ray et al., 2014; Tang et al., 2014a, 2015, 2016). Detailed descrip- 100 ms. In the four cell groups, there were only a few cells where the spikes tions of recording procedures (Pinault, 1996; Lee et al., 2006; Herfst et al., did not have sufficient quality to analyze the spike shape, and we could analyze 2012; Tang et al., 2014b), quality control (Joshua et al., 2007), tissue prepara- 19/22 parasubicular cells, 24/31 MEC L2 pyramidal cells, 58/68 MEC L2 stel- tion, immunohistochemistry, and image acquisition (Naumann et al., 2016; Ray late cells, and 27/32 MEC L3 cells. We calculated the mean spike shape of and Brecht, 2016), can be found in these papers and in the Supplemental every cell and determined the spike features from these traces. For plotting Experimental Procedures. the comparison between cells and for illustrating the differences in peak-to- trough time (Figures 4A and 4B), we normalized the spike shape by subtracting Classification of Non-identified Layer 2 Neurons the noise floor, dividing the mean spike by the peak-to-trough height, and In addition to labeled cells, we included a number of unlabeled, regularly setting the peak height to 1. spiking cells from MEC L2 in our analysis. These cells were assigned as either putatively calbindin-positive (pCb+) pyramidal cells or putatively calbindin- Analysis of Theta Rhythmicity and Theta Cycle Skipping negative (pCb ) stellate cells based on their theta strength and preferred theta To determine whether a neuron was rhythmic and theta cycle-skipping, we À phase angle using the classification approach of Tang et al. (2014a); i.e., based used an MLE of a parametric model of the ISI histogram (‘‘mle_rhythmicity’’; on the theta strength and preferred theta phase angle of spiking activity. As in Climer et al., 2015). For every cell, we fitted three models to the ISI distribution: Tang et al. (2014a), we used a 0.1 guard zone and found that the cells were well a flat model with no rhythmic components, a rhythmic, non-skipping model with separated with no cells in the guard zone (Figure S1). In the manuscript, we a rhythmic modulation of the ISI histogram, and a rhythmic, cycle-skipping

Cell Reports 16, 1005–1015, July 26, 2016 1013

76 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

model with a rhythmic modulation of the ISI histogram and a second parameter istry and microscopy. S.S., R.K., and M.B. provided expertise and feedback on introducing theta cycle skipping (i.e., a higher amplitude of every other peak in the analysis and supervised the project. C.L.E. conceived the study and wrote the ISI histogram). When fitting the models, we searched for a rhythmic compo- the first version of the manuscript. All authors provided feedback and contrib- nent with a theta frequency between 5 and 13 Hz and for cycle skippings >0.01. uted to writing the manuscript. The three fitted models were compared using the appropriate c2 statistic (calculated from the maximum log likelihood of the models) to generate two ACKNOWLEDGMENTS

p values: prhythmic (comparing the ﬂat and the rhythmic, non-skipping models)

and pskipping (comparing the rhythmic, non-skipping and the rhythmic, cycle- This work was supported by the Humboldt Universita¨ t zu Berlin, BCCN Berlin skipping models). The cells were classiﬁed using a two-level classiﬁcation (Fig- (German Federal Ministry of Education and Research BMBF, Fo¨ rderkennzei-

ure 5B). First we determined whether a cell was rhythmic (prhythmic < 0.05) chen 01GQ1001A, 01GQ0901, 01GQ1403, and 01GQ0972), NeuroCure, the

or non-rhythmic (prhythmic > 0.05). Then we classiﬁed the rhythmic cells as Neuro-Behavior ERC grant, and the Gottfried Wilhelm Leibniz Prize of the

either rhythmic, cycle-skipping (pskipping < 0.05) or rhythmic, non-skipping DFG. We thank Andreea Neukirchner, Juliane Steger, and Undine Schneeweiß

(pskipping > 0.05). for technical assistance. To statistically assess whether theta cycle skipping cells were rarer among rhythmic cells in the generally non-rhythmic cell types (MEC L2 stellates and Received: March 30, 2016 MEC L3 neurons) than in the generally rhythmic cell types (parasubicular neurons Revised: May 4, 2016 and MEC L2 pyramids), we fitted a mixed-effects logistic regression. We con- Accepted: June 12, 2016 structed a vector, isGenRhytm (which takes the value1 for parasubicular neurons Published: July 14, 2016 and MEC L2 pyramids and the value 0 for MEC L2 stellates and MEC L3 neurons). We also constructed a vector type that simply dummy-coded the four neuron REFERENCES types from 1, 2, 3, and 4. We dummy-coded when the neuron was theta cycle- skippingin the vector isSkipping. We then modeled the probability of being rhyth- Aarts, E., Verhage, M., Veenvliet, J.V., Dolan, C.V., and van der Sluis, S. (2014). mic as a function of being generally rhythmic while controlling for the different A solution to dependency: using multilevel analysis to accommodate nested number of cells in the four categories of neurons: ‘‘isSkipping isGenRhytm + data. Nat. Neurosci. 17, 491–496. � (1 type)’’ using ‘‘fitglme’’ in MATLAB (Aarts et al., 2014). Akaike, H. (1974). A new look at the statistical model identification. IEEE Trans. j In addition to the MLE approach, we also calculated the theta strength and Automat. Contr. 19, 716–723. preferred theta phase of every cell. The local field potential was bandpass- Alessi, C., Raspanti, A., and Magistretti, J. (2016). Two Distinct Types of Depo- filtered in the theta range (4–12 Hz), and a Hilbert transform was used to deter- larizing Afterpotentials are Differentially Expressed in Stellate and Pyramidal- mine the instantaneous phase of the theta wave for every spike. The theta lock- like Neurons of Entorhinal-Cortex Layer II. Hippocampus. ing strength and the preferred phase angle were calculated as the modulus and Alonso, A., and Klink, R. (1993). Differential electroresponsiveness of stellate argument of the Rayleigh average vector of the theta phase at all spike times. and pyramidal-like cells of medial entorhinal cortex layer II. J. Neurophysiol. 70, 128–143. Statistical Modeling Boccara, C.N., Sargolini, F., Thoresen, V.H., Solstad, T., Witter, M.P., Moser, Statistical modeling (generalized linear models) was done in MATLAB using E.I., and Moser, M.B. (2010). Grid cells in pre- and parasubiculum. Nat. Neuro- the ‘‘glmefit’’ function. We modeled burstiness as a function of theta strength sci. 13, 987–994. as a normally distributed variable (Figures S3A–S3C). We modeled the peak- to-trough time as a gamma-distributed variable with a reciprocal link function Botev, Z.I., Grotowaki, J.F., and Kroese, D.P. (2010). Kernel density estimation in MATLAB because it can only assume positive values (Figures S4A–S4C). To via diffusion. Ann. Stat. 38, 2916–2957. compare models, we either calculated and compared the AIC (Akaike, 1974) Brandon, M.P., Bogaard, A.R., Schultheiss, N.W., and Hasselmo, M.E. (2013). or, in the case of nested models, calculated the p value from likelihood ratio Segregation of cortical head direction cell assemblies on alternating q cycles. tests. In the manuscript, we describe all statistical models using standard Nat Neurosci. 16, 739–748. Wilkinson notation (Wilkinson and Rogers, 1973). Brecht, M., Ray, S., Burgalossi, A., Tang, Q., Schmidt, H., and Naumann, R. (2013). An isomorphic mapping hypothesis of the grid representation. Philos. Analysis of Phase Precession Trans. R. Soc. Lond. B Biol. Sci. 369, 20120521. To identify coherent periods of elevated firing (‘‘single runs’’), we follow a pre- Burgalossi, A., Herfst, L., von Heimendahl, M., Fo¨ rste, H., Haskic, K., Schmidt, viously applied strategy based on the temporal structure of the recorded spike M., and Brecht, M. (2011). Microcircuits of functionally identified neurons in the trains (Reifenstein et al., 2016). Briefly, we convolved the spike train with a rat medial entorhinal cortex. Neuron 70, 773–786. Gaussian kernel to estimate the instantaneous firing rate. We then used a firing Burgess, N. (2014). The 2014 Nobel Prize in Physiology or Medicine: a spatial rate threshold to locate periods of elevated firing (Figure 6A, top). For each of model for cognitive neuroscience. Neuron 84, 1120–1125. the single runs, the times and theta phases of all spikes were used to assess Climer, J.R., DiTullio, R., Newman, E.L., Hasselmo, M.E., and Eden, U.T. phase precession. We quantified phase precession by calculating the slope, (2015). Examination of rhythmicity of extracellularly recorded neurons in the phase range, and circular-linear correlation coefficient of the circular-linear entorhinal cortex. Hippocampus 25, 460–473. regression line (Figure 6A, bottom; Kempter et al., 2012; Reifenstein et al., 2012, 2014, 2016). Csicsvari, J., Hirase, H., Czurko´ , A., Mamiya, A., and Buzsa´ ki, G. (1999). Oscillatory coupling of hippocampal pyramidal cells and interneurons in the behaving Rat. J. Neurosci. 19, 274–287. SUPPLEMENTAL INFORMATION Domnisoru, C., Kinkhabwala, A.A., and Tank, D.W. (2013). Membrane potential dynamics of grid cells. Nature 495, 199–204. Supplemental Information includes Supplemental Experimental Procedures and five figures and can be found with this article online at http://dx.doi.org/ Fuchs, E.C., Neitz, A., Pinna, R., Melzer, S., Caputi, A., and Monyer, H. (2016). 10.1016/j.celrep.2016.06.057. Local and Distant Input Controlling Excitation in Layer II of the Medial Entorhi- nal Cortex. Neuron 89, 194–208. AUTHOR CONTRIBUTIONS Giocomo, L.M., Moser, M.-B., and Moser, E.I. (2011). Computational models of grid cells. Neuron 71, 589–603. C.L.E. analyzed burstiness, spike shape, theta rhythmicity, and cycle skipping. Glasgow, S.D., and Chapman, C.A. (2008). Conductances mediating intrinsic E.T.R. analyzed phase precession. Q.T. and A.B. provided access to juxtacel- theta-frequency membrane potential oscillations in layer II parasubicular neu- lular data and assisted with data analysis. S.R. performed immunohistochem- rons. J. Neurophysiol. 100, 2746–2756.

1014 Cell Reports 16, 1005–1015, July 26, 2016

77 Ebbesen (2017) Chapter 6

Gold, C., Girardin, C.C., Martin, K.A., and Koch, C. (2009). High-amplitude Ray, S., and Brecht, M. (2016). Structural development and dorsoventral matu- positive spikes recorded extracellularly in cat visual cortex. J. Neurophysiol. ration of the medial entorhinal cortex. eLife 5, 13343. 102, 3340–3351. Ray, S., Naumann, R., Burgalossi, A., Tang, Q., Schmidt, H., and Brecht, M. Hafting, T., Fyhn, M., Molden, S., Moser, M.B., and Moser, E.I. (2005). Micro- (2014). Grid-layout and theta-modulation of layer 2 pyramidal neurons in structure of a spatial map in the entorhinal cortex. Nature 436, 801–806. medial entorhinal cortex. Science 343, 891–896. Hafting, T., Fyhn, M., Bonnevie, T., Moser, M.B., and Moser, E.I. (2008). Hip- Reifenstein, E.T., Kempter, R., Schreiber, S., Stemmler, M.B., and Herz, A.V. pocampus-independent phase precession in entorhinal grid cells. Nature (2012). Grid cells in rat entorhinal cortex encode physical space with indepen- 453, 1248–1252. dent firing fields and phase precession at the single-trial level. Proc. Natl. Herfst, L., Burgalossi, A., Haskic, K., Tukker, J.J., Schmidt, M., and Brecht, M. Acad. Sci. USA 109, 6301–6306. (2012). Friction-based stabilization of juxtacellular recordings in freely moving Reifenstein, E., Stemmler, M., Herz, A.V.M., Kempter, R., and Schreiber, S. rats. J. Neurophysiol. 108, 697–707. (2014). Movement dependence and layer specificity of entorhinal phase pre- Jaramillo, J., Schmidt, R., and Kempter, R. (2014). Modeling inheritance of cession in two-dimensional environments. PLoS ONE 9, e100638. phase precession in the hippocampal formation. J. Neurosci. 34, 7715–7731. Reifenstein, E.T., Ebbesen, C.L., Tang, Q., Brecht, M., Schreiber, S., and Jeewajee, A., Barry, C., Douchamps, V., Manson, D., Lever, C., and Burgess, Kempter, R. (2016). Cell-Type Specific Phase Precession in Layer II of the N. (2013). Theta phase precession of grid and place cell firing in open environ- Medial Entorhinal Cortex. J. Neurosci. 36, 2283–2288. ments. Philos. Trans. R Soc. Lond. B Biol. Sci. 369, 20120532. Sargolini, F., Fyhn, M., Hafting, T., McNaughton, B.L., Witter, M.P., Moser, Joshua, M., Elias, S., Levine, O., and Bergman, H. (2007). Quantifying the isola- M.B., and Moser, E.I. (2006). Conjunctive representation of position, direction, tion quality of extracellularly recorded action potentials. J. Neurosci. Methods and velocity in entorhinal cortex. Science 312, 758–762. 163, 267–282. Schmidt, R., Diba, K., Leibold, C., Schmitz, D., Buzsa´ ki, G., and Kempter, R. Kempter, R., Leibold, C., Buzsa´ ki, G., Diba, K., and Schmidt, R. (2012). Quanti- (2009). Single-trial phase precession in the hippocampus. J. Neurosci. 29, fying circular-linear associations: hippocampal phase precession. J. Neurosci. 13232–13241. Methods 207, 113–124. Sheffield, M.E.J., and Dombeck, D.A. (2015). The binding solution? Nat. Neu- Kinkhabwala AA, Aronov D, Tank DW (2015) Visual cue-related activity of MEC rosci. 18, 1060–1062. cells during navigation in virtual reality. Program No. 632.21/CC2. 2015 Neuro- science Meeting Planner. Washington, DC: Society for Neuroscience, 2015. Solstad, T., Boccara, C.N., Kropff, E., Moser, M.B., and Moser, E.I. (2008). Online. Representation of geometric borders in the entorhinal cortex. Science 322, Kitamura, T., Pignatelli, M., Suh, J., Kohara, K., Yoshiki, A., Abe, K., and Tone- 1865–1868. gawa, S. (2014). Island cells control temporal association memory. Science Sun, C., Kitamura, T., Yamamoto, J., Martin, J., Pignatelli, M., Kitch, L.J., 343, 896–901. Schnitzer, M.J., and Tonegawa, S. (2015). Distinct speed dependence of ento- Koenig, J., Linder, A.N., Leutgeb, J.K., and Leutgeb, S. (2011). The spatial perirhinal island and ocean cells, including respective grid cells. Proc. Natl. Acad. odicity of grid cells is not sustained during reduced theta oscillations. Science Sci. USA 112, 9466–9471. 332, 592–595. Tang, Q., Burgalossi, A., Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Kropff, E., Carmichael, J.E., Moser, M.B., and Moser, E.I. (2015). Speed cells in Spicher, D., and Brecht, M. (2014a). Pyramidal and stellate cell specificity of the medial entorhinal cortex. Nature 523, 419–424. grid and border representations in layer 2 of medial entorhinal cortex. Neuron Latuske, P., Toader, O., and Allen, K. (2015). Interspike Intervals Reveal Func- 84, 1191–1197. tionally Distinct Cell Populations in the Medial Entorhinal Cortex. J. Neurosci. Tang, Q., Brecht, M., and Burgalossi, A. (2014b). Juxtacellular recording 35, 10963–10976. and morphological identification of single neurons in freely moving rats. Nat. Lee, A.K., Manns, I.D., Sakmann, B., and Brecht, M. (2006). Whole-cell record- Protoc. 9, 2369–2381. ings in freely moving rats. Neuron 51, 399–407. Tang, Q., Ebbesen, C.L., Sanguinetti-Scheck, J.I., Preston-Ferrer, P., Gundl- Lever, C., Burton, S., Jeewajee, A., O’Keefe, J., and Burgess, N. (2009). finger, A., Winterer, J., Beed, P., Ray, S., Naumann, R., Schmitz, D., et al. Boundary vector cells in the subiculum of the hippocampal formation. (2015). Anatomical Organization and Spatiotemporal Firing Patterns of Layer J. Neurosci. 29, 9771–9777. 3 Neurons in the Rat Medial Entorhinal Cortex. J. Neurosci. 35, 12346–12354. Mizuseki, K., Sirota, A., Pastalkova, E., and Buzsa´ ki, G. (2009). Theta oscilla- Tang, Q., Burgalossi, A., Ebbesen, C.L., Sanguinetti-Scheck, J.I., Schmidt, H., tions provide temporal windows for local circuit computation in the entorhi- Tukker, J.J., Naumann, R., Ray, S., Preston-Ferrer, P., Schmitz, D., and nal-hippocampal loop. Neuron 64, 267–280. Brecht, M. (2016). Functional Architecture of the Rat Parasubiculum. Mountcastle, V.B., Talbot, W.H., Sakata, H., and Hyva¨ rinen, J. (1969). Cortical J. Neurosci. 36, 2289–2301. neuronal mechanisms in flutter-vibration studied in unanesthetized mon- Unal, G., Joshi, A., Viney, T.J., Kis, V., and Somogyi, P. (2015). Synaptic Tar- keys. Neuronal periodicity and frequency discrimination. J. Neurophysiol. gets of Medial Septal Projections in the Hippocampus and Extrahippocampal 32, 452–484. Cortices of the Mouse. J. Neurosci. 35, 15812–15826. Naumann, R.K., Ray, S., Prokop, S., Las, L., Heppner, F.L., and Brecht, M. Varga, C., Lee, S.Y., and Soltesz, I. (2010). Target-selective GABAergic control (2016). Conserved size and periodicity of pyramidal patches in layer 2 of of entorhinal cortex output. Nat. Neurosci. 13, 822–824. medial/caudal entorhinal cortex. J. Comp. Neurol. 524, 783–806. Welday, A.C., Shlifer, I.G., Bloom, M.L., Zhang, K., and Blair, H.T. (2011). Newman, E.L., and Hasselmo, M.E. (2014). Grid cell firing properties vary as a Cosine directional tuning of theta cell burst frequencies: evidence for spatial function of theta phase locking preferences in the rat medial entorhinal cortex. coding by oscillatory interference. J. Neurosci. 31, 16157–16176. Front. Syst. Neurosci. 8, 193. Wilkinson, G.N., and Rogers, C.E. (1973). Symbolic Description of Factorial Pastoll, H., Ramsden, H.L., and Nolan, M.F. (2012). Intrinsic electrophysiolog- Models for Analysis of Variance. J. R. Stat. Soc. Ser. C Appl. Stat. 22, 392–399. ical properties of entorhinal cortex stellate cells and their contribution to grid cell firing fields. Front. Neural Circuits 6, 17. Winter, S.S., Clark, B.J., and Taube, J.S. (2015). Spatial navigation. Disruption Pinault, D. (1996). A novel single-cell staining procedure performed in vivo un- of the head direction cell network impairs the parahippocampal grid cell signal. der electrophysiological control: morpho-functional features of juxtacellularly Science 347, 870–874. labeled thalamic cells and other central neurons with biocytin or Neurobiotin. Zilli, E.A. (2012). Models of grid cell spatial firing published 2005-2011. Front. J. Neurosci. Methods 65, 113–136. Neural Circuits 6, 16.

Cell Reports 16, 1005–1015, July 26, 2016 1015

78 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Cell Reports, Volume 16

Supplemental Information

Cell Type-Speciﬁc Differences in Spike Timing and Spike Shape in the Rat Parasubiculum and Superﬁcial Medial Entorhinal Cortex

Christian Laut Ebbesen, Eric Torsten Reifenstein, Qiusong Tang, Andrea Burgalossi, Saikat Ray, Susanne Schreiber, Richard Kempter, and Michael Brecht

79 Ebbesen (2017) Chapter 6

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 1

Figure S1: Classification of medial entorhinal cortex layer 2 neurons, Related to Figure 1.

(A) Polar plot of theta strength and theta phase angle, φ, of the spiking activity of calbindin+

pyramidal cells (Pyr, green dots) and calbindin- stellate cells (Stel, black dots) identified in

freely moving rats. Lines indicate mean theta phase angle and median theta strength.

(B) Polar plot of theta strength and preferred theta phase angle, φ, for nonidentified MEC L2

cells, which were classified as putative calbindin+ cells (pCb+, white dots) and putative

calbindin- cells (pCb-, dark grey dots). The background colors indicate the two classification

groups (light green and light grey for putative calbindin+ (pCb+) and putative calbindin-

(pCb-)) and the guard zone around the classification boundary (white). No nonidentified cell

fell within the guard zone.

80 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Supplemental , Related to Figure , 3. Related

type specific differences in burstiness differences specific type - MEC Cell Type, Spike Shape & Discharge Timing Shape Spike Cell MEC Type, &

(Legend on next page) on next (Legend Figure S3: Classification of medial entorhinal cortex layer 2 cells does not explain cell explain not does layer 2 cells cortex entorhinal of medial Classification Figure S3: Ebbesen et al. et Ebbesen

81 Ebbesen (2017) Chapter 6

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 3

Figure S3: Classification of medial entorhinal cortex layer 2 cells does not explain cell-type specific differences in burstiness, Related to Figure 3

(A) Graphical representation of three fitted generalized linear models to investigate if burstiness is

related to theta strength (Model 1), putative cell type (Model 2) or both (Model 3).

(B) Comparison of the Akaike information criterion (‘AIC’) of the three models. P-values indicate

theoretical likelihood ratio tests between nested models. Both the AIC and the likelihood ratio

tests suggest that Model 2 is superior to the other models.

relationship between theta strength and burstiness (PStrength = 0.00015), but this effect disappears

when we add cell type as another independent variable (Model 3: PType = 0.014, PStrength = 0.54).

The best model depends only on cell type (Model 2: PType = 0.0000076).

(D) Left: Comparison of the burstiness for the four different neuron types as in Figure 3C, i.e.

including MEC L2 cells classified as putatively pyramidal and putatively stellates (Same as Fig

3C). Right: Same plot, but only including identified MEC L2 cells. P-values indicate results of

t-tests (assuming equal (Cb- vs. L3) or unequal (PaS, Cb+ vs. Cb-, L3) variances). Due to the

low number of Cb+ cells, we report the P-value both with and without the statistical outlier

(indicated by arrow).

82 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 4

Figure S4. Spike shape is a feature of cell type rather than burstiness, Related to Figure 4.

(A) Scatterplot of the peak-to-trough time as a function of burstiness for the four neuron types

defined in Fig. 4A; symbols indicated by legend. Lines indicate the estimated peak-to-trough

times as a function of burstiness and cell type from the generalized linear model 1 (GLM1,

dashed grey line) and GLM2 (best model, four solid horizontal lines, color code as for symbols).

(B) Comparison of degrees of freedom (“params”) and Akaike information criterion (AIC) for four

GLMs in (A). P-values indicate theoretical likelihood ratio tests for nested models. According to

the AIC, the best model is GLM2 (Peak-to-trough time depends only on neuron type, not on

burstiness). Similarly, a likelihood ratio test rejects the inclusion of burstiness as an extra

predictor variable (p = 0.32 for GLM2 vs. GLM3). There is no indication of a statistical

interaction between cell type and burstiness (GLM4).

83 Ebbesen (2017) Chapter 6

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 5

and burstiness.

84 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 6

Figure S5. Comparison of rhythmicity and cycle-skipping with and without including classified

medial entorhinal cortex layer 2 cells show similar patterns, Related to Figure 5.

(A) Left: Same plot as Figure 5C, proportions (above) and counts (below) of non-rhythmic,

rhythmic and cycle-skipping neurons among the cell types, i.e. including MEC L2 cells

classified as putatively pyramidal and putatively stellate. Right: Same plot, but only including

identified MEC L2 cells. (P values indicate χ2 tests of equal proportions among cell types).

85 Ebbesen (2017) Chapter 6

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 7

Supplemental experimental procedures

Juxtacellular recordings

In this paper, we analyzed a data set of juxtacellular recordings from the superficial medial entorhinal

cortex and the parasubiculum which we have previously published (Ray et al. 2014, Tang et al. 2014,

Tang et al. 2015, Tang et al. 2016). Below, we present a summary of the recording procedure from

these previous papers.

Juxtacellular recordings and tetrode recordings in freely moving animals were obtained in male Wistar

and Long-Evans rats (150-250 g). Experimental procedures were essentially performed as recently

described (Tang et al., 2014a; Tang et al., 2014b). Briefly, for juxtacellular recordings, glass pipettes

with resistance 4-6 MΩ were filled with extracellular (Ringer) solution containing (in mM) NaCl 135,

KCl 5.4, HEPES 5, CaCl2 1.8, and MgCl2 1 (pH = 7.2) and Neurobiotin (1-2%). The glass recording

pipette was advanced into the brain by means of a miniaturized micromanipuator (Tang et al 2014b)

while rats explored open field arenas (70 x 70 cm or 1 x 1 m square black box, with a white cue card

on the wall). Juxtacellular labeling was attempted at the end of the recording session according to

standard procedures (Pinault, 1996). Unidentified recordings in parasubiculum and MEC were either

lost before the labeling could be attempted, or the recorded neurons could not be unequivocally

identified, as described in Tang et al., 2014a, Tang et al 2015, Tang et al., 2016. After the experiment,

the animals were euthanized with an overdose of ketamine, urethane or pentobarbital, and perfused

transcardially with 0.1 M phosphate buffer followed by 4% paraformaldehyde solution. The

juxtacellular signals were amplified by the ELC-03XS amplifier (NPI Electronics, Tamm, Germany)

and sampled at 20 kHz by a data-acquisition interface under the control of PatchMaster 2.20 software

(HEKA, Ludwigshafen, Germany). The animal’s location and head-direction was automatically

tracked at 25 Hz by the Neuralynx video tracking system and two head-mounted LEDs.

Tissue preparation, immunohistochemistry, and image acquisition

Rats were anaesthetized by isoflurane and euthanized by an intraperitoneal injection of 20% urethane.

Animals were then transcardially perfused with 0.9% phosphate-buffered saline, followed by PFA.

86 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 8

Subsequently, brains were removed from the skull and postfixed in PFA overnight. Brains were then

immersed in 10% sucrose and then in 30% sucrose for at least one night for cryoprotection. The

brains were embedded in Jung tissue Freeing Medium (Leica Microsystems Nussloch, Germany), and

mounted on a freezing microtome (Leica 2035 Biocut) to obain tangential and parasaggital sections at

60 microns.

Tangential sections of the medial entorhinal cortex and parasubiculum were obtained as previously

described (Ray et al., 2014; Naumann et al., 2016) by separating the entorhinal cortex from the

remaining hemisphere by a cut parallel to the face of the medial entorhinal cortex (Ray & Brecht,

2016) and sectioning with the surface of the entorhinal cortex attached to the block face of the

microtome.

Immunohistochemical stains were performed on tangential and sagittal sections. The sections were

pre-incubated in a blocking solution containing 0.1 M PBS, 2% Bovine Serum Albumin (BSA) and

0.5% Triton X-100 (PBS-X) for an hour at room temperature (RT). Following this, primary antibodies

were diluted in a solution containing PBS-X and 1% BSA. Primary antibodies against the calcium

binding proteins Calbindin (Swant: CB300, CB 38; 1:5000), the transmembrane protein Wolframin

(Proteintech: 11558-1-AP; 1:200), and the calmodulin binding protein Purkinje cell protein 4 (Sigma:

HPA005792; 1:200) were used. Sections were incubated in primary antibodies for at least 24 hours

under mild shaking at 4 degrees centigrade. Subsequently sections were incubated in secondary

antibodies coupled to different fluorophores (Alexa 488, 546; Invitogen; 1:500). For multiple antibody

labelling, antibodies raised in different host species were used.

Images were acquired with a Leica DM5500B epifluorescence microscope with a Leica DFC345 FX

camera. Alexa fluorophores were excited using the appropriate filters (Alexa 488- L5; Alexa 546- N3).

Fluorescent images were acquired in monochrome, and color maps were applied to the images post

acquisition. Post hoc linear brightness and contrast adjustment were applied uniformly to the image

under analysis.

87 Ebbesen (2017) Chapter 6

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 9

Quality control across juxtacellular recordings

We checked explicitly for systematic differences in LFP power across recordings. First, we calculated

the global theta power in the LFP of all recordings, defined as the mean power spectral density of the

theta-peak in the LFP spectrogram ± 0.3 Hz. We did not find any significant differences in LFP theta

power among cell types (P > 0.05, one-way ANOVA). We also did not find any correlation between

the LFP theta power and burstiness in our data (P > 0.05, Spearman correlation).

If spikes are missed during bursts, this would bias a recording towards low burstiness. Under the

juxtacellular recording configuration, however, spikes are well above the noise level (signal-to-noise

typically an order of magnitude higher than tetrode recordings) and thus unlikely to fall below the

detection threshold. It is the case, however, juxtacellular recordings might potentially be more

disruptive for the recorded neurons due to the close proximity of the glass tip and the membrane;

recordings (or portions of recordings) where signs of cellular damage were observed (e.g. action-

potential broadening, increase in firing rate; see Pinault et al., 1996; Herfst et al., 2012) were excluded

from the analysis. As a measure of ‘recording quality’, we estimated the signal-to-noise ratio of the

spikes (Joshua et al. 2007), and we found no difference between cell types (P > 0.05, Kruskal-Wallis

test). We also found no correlation between recording quality and burstiness, spike shape or phase

precession (all P > 0.05, Spearman correlations).

88 Ebbesen (2017) MEC Cell Type, Spike Shape & Discharge Timing

Ebbesen et al. MEC Cell Type, Spike Shape & Discharge Timing Supplemental 10

Supplemental references

Herfst L, Burgalossi A, Haskic K, Tukker JJ, Schmidt M, Brecht M. (2012) Friction-based stabilization of juxtacellular recordings in freely moving rats. J Neurophysiol. (2):697-707.

Joshua M, Elias S, Levine O, Bergman H. (2007) Quantifying the isolation quality of extracellularly recorded action potentials. J Neurosci Methods. 163(2):267-82.

Naumann, RK, Ray, S, Prokop, S, Las, L, Heppner, FL, Brecht M. (2016) Conserved size and periodicity of pyramidal patches in layer 2 of medial/caudal entorhinal cortex. J Comp. Neurol. 524:783-806

Pinault D. (1996) A novel single-cell staining procedure performed in vivo under electrophysiological control: morpho-functional features of juxtacellularly labeled thalamic cells and other central neurons with biocytin or Neurobiotin. J Neurosci Methods; 65(2):113-36.

Ray, S., Naumann, R., Burgalossi, A., Tang, Q., Schmidt, H., Brecht, M. (2014) Grid-layout and theta- modulation of layer 2 pyramidal neurons in medial entorhinal cortex. Science 343, 891–6.

Ray S, Brecht M (2016) Structural development and dorsoventral maturation of the medial entorhinal cortex. eLife. e13343.

Tang, Q., Burgalossi, A., Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Spicher, D., Brecht, M. (2014a) Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex. Neuron 84, 1191–1197.

Tang, Q., Brecht, M., Burgalossi, A. (2014b) Juxtacellular recording and morphological identification of single neurons in freely moving rats. Nat Protoc. 9(10), 2369-81.

Tang, Q., Ebbesen, C.L., Sanguinetti-Scheck, J.I., Preston-Ferrer, P., Gundlfinger, A., Winterer, J., Beed, P., Ray, S., Naumann, R., Schmitz, D., Brecht, M., Burgalossi, A. (2015) Anatomical Organization and Spatiotemporal Firing Patterns of Layer 3 Neurons in the Rat Medial Entorhinal Cortex. J. Neurosci. 35(36), 12346–12354.

Tang, Q., Burgalossi, A., Ebbesen, C.L., Sanguinetti-Scheck, J.I., Schmidt, H., Tukker, J.J., Naumann, R., Ray, S., Preston-Ferrer, P., Schmitz, D., Brecht, M. (2016) Functional Architechture of the Rat Parasubiculum. J. Neurosci. 36(7):2289-301

89 Ebbesen (2017) Chapter 7 7. Vibrissa motor cortex activity suppresses contralateral whisking behavior

This manuscript was published as: Ebbesen, C.L., Doron, G., Lenschow, C., & Brecht, M. (2017). Vibrissa motor cortex activity suppresses contralateral whisking behavior. Nature Neuroscience 20(1):82-89 This is the author’s version of this work, reprinted with permission from Nature Pub. Group

Figure 10: Our suggested cover image, based on the article: A Wistar rat it reins. The reins symbolize the suppressive effect of motor cortex, which is “reining in” the behavior of the rat. The rat, grass and flowers is a 3d model by Shimpei Ishiyama.

A preliminary subset of the data presented in this article were shown in my M.Sc. thesis (2013).

90 Ebbesen (2017) VMC activity suppresses contralateral touch

4 Vibrissa motor cortex activity suppresses

5 contralateral whisker touch

7 by

9 Christian Laut Ebbesen(1,2), Guy Doron(1,3), Constanze Lenschow(1) and Michael Brecht(1).

10 (1)Bernstein Center for Computational Neuroscience Berlin, Humboldt-Universität zu Berlin, 11 Berlin, Germany

12 (2)Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Berlin, Germany

13 (3)Current Address: NeuroCure Cluster of Excellence, Humboldt-Universität zu Berlin, Berlin, 14 Germany

91 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 2

20 Anatomical1,2, stimulation1–8 and lesion data9,10 point to a role of vibrissa motor cortex in 21 the control of whisker movement. Motor cortex is classically thought to play a key role 22 in movement generation11–13, but most studies have found only weak correlations be- 23 tween vibrissa motor cortex activity and whisking14–17. The exact role of vibrissa motor 24 cortex in motor control remains unknown. To address this question we recorded vibris- 25 sa motor cortex neurons during various forms of vibrissal touch, all of which were asso- 26 ciated with increased movement and forward positioning of whiskers. Free whisking, 27 palpation of objects and social touch all resulted in similar vibrissa motor cortex re- 28 sponses: (i) Population activity decreased. (ii) The vast majority (~80%) of significantly 29 modulated single cells decreased their firing. (iii) Rate-decreasing cells were the most 30 strongly modulated cells. To understand the cellular basis of this decrease of activity, we 31 performed juxtacellular recordings, nanostimulation and in vivo whole-cell recordings in 32 head-fixed animals. Social facial touch – a strongly engaging stimulus18–20 – resulted in 33 decreased spiking activity, decreased cell excitability and a ~1.5 mV hyperpolarization 34 in vibrissa motor cortex neurons. High-speed videography and generalized linear model- 35 ing of the spiking patterns of identified deep layer output neurons during social facial 36 touch episodes revealed that the observed suppression of VMC activity is likely due to 37 nose-to-nose touch and whisker protraction. To assess how activation of vibrissa motor 38 cortex impacts whisking behavior we performed intra-cortical microstimulation, which 39 led to whisker retraction, as if to abort vibrissal touch. Finally, we blocked vibrissa mo- 40 tor cortex. A variety of inactivation protocols resulted in an increase of contralateral 41 whisker movements and contralateral whisker protraction, as if to engage in vibrissal 42 touch. These observations suggest that the role of vibrissa motor cortex is not restricted 43 to movement generation. Instead, the data collectively point to movement suppression as 44 a prime function of vibrissa motor cortex activity.

92 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 3

46 Vibrissa motor cortex (VMC, Fig 1a) is a cortical vibrissa representation originally identified 47 by a variety of stimulation techniques1–8. The huge size of this representation possibly reflects 48 the great ecological relevance of vibrissa movements for rats21,22. In contrast to classic studies 49 on primate primary motor cortex (M1) activity11,12, VMC population activity is only weakly 50 correlated with movement14–17. It is not entirely clear why the correlation between whisker 51 movement and VMC activity is weak, but we note that most of what we know about VMC 52 activity during whisking comes from recordings in animals simply whisking in air14–17. Stud- 53 ies on primate motor cortex have shown, that besides the musculotopic representation of body 54 movements12,13, the motor cortex might also represent a map of ecologically relevant behav- 55 iors23. The information about VMC activity during self-initiated, ecologically relevant behav- 56 iors is still limited and it remains unclear how VMC contributes to motor control during such 57 behaviors. This prompted us to pose the following questions about VMC function: (1) How is 58 the activity of the VMC “output layers”1 modulated, when rats engage in various ecologically 59 relevant whisking behaviors? (2) What are the cellular mechanisms, which contribute to the 60 modulation of VMC activity? (3) How does an increase of VMC activity by microstimulation 61 during whisking affect ongoing whisking movements? (4) How does a decrease of VMC ac- 62 tivity by pharmacological blockade affect whisking movements?

93 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 4

64 RESULTS:

65 Vibrissa motor cortex firing decreases during various forms of vibrissal touch

66 We investigated VMC modulation by three self-initiated rat whisker behaviors (Fig. 1b): free 67 whisking (explorative whisking bouts in air), object touch (whisking onto objects) and social 68 touch (whisking onto conspecifics)19. All whisker-behaviors were compared to rest (animal 69 not whisking). Single unit activity was recorded from VMC layer 5 using tetrodes. With high- 70 speed videography, we quantified the whisker set angle and whisking power during the vari- 71 ous behaviors (Fig. 1c). We found that during all whisker behaviors the whiskers were held at 72 a more protracted set angle than at rest, on average by 19° (Fig. 1d top, P < 0.001, one-way 73 ANOVA, all P < 0.001, unpaired t-tests). By definition, during all whisking behaviors, the 74 whisking power was higher than at rest (Fig. 1d bottom).

75 In Fig. 1e, we show a raster plot and a peri-stimulus time histogram (PSTH) of an example 76 layer 5 cell aligned to the beginning of free whisking. The PSTH shows the predominant re- 77 sponse pattern: a decrease in firing rate during free whisking. We observed a large variability 78 of responses (Fig. 1e, top). Some cells increased their firing, some were not modulated and 79 some decreased their firing rate, but as a whole the population activity was significantly de- 80 creased during free whisking (Fig. 1f top, median 2.31/2.05 Hz Baseline/free whisking, Slope 81 = 0.812, P = 0.00010, N = 158 cells, Mann-Whitney U-test). We assessed the significance of 82 firing rate changes by a bootstrapping procedure and found that 80% of significantly modulat- 83 ed cells decreased their activity in free whisking (P = 0.000041, two-tailed binomial test for 84 equal proportions). We restricted our analysis to cells with a firing rate below 10 Hz to reduce 85 the proportion of interneurons (see Methods). In the small subset of cells with firing rates > 86 10 Hz (14% of cells) we found no significant rate changes (All P > 0.05, Mann-Whitney U- 87 test). Inclusion of high-firing cells did not change the results. To quantify the modulation of 88 single cells, we calculated a modulation index (see Methods) and found that the most strongly 89 modulated cells were the cells that decreased their firing rate (Fig. 1f bottom, P < 0.05, Mann- 90 Whitney U-test). We wondered, if the firing rate decrease is also to be seen in more challeng- 91 ing forms of vibrissal touch. For both object touch (Fig. 1g,h top, median 2.20/1.65 Hz Base- 92 line/Touch, Slope = 0.749, P = 0.023, N = 122 cells, Mann-Whitney U-test) and for social 93 facial touch (Fig. 1i,j top, median 2.26/1.87 Hz Baseline/Touch, Slope = 0.806 P = 0.00018, N 94 = 156 cells, Mann-Whitney U-test), neurons also decreased their firing rate. Specifically, we 95 observed a decrease in 88% of the cells significantly modulated by object touch (Fig. 1h bot- 96 tom) and in 78% of the cells significantly modulated by social touch, (Fig. 1j bottom, both P < 97 0.05, two-tailed binomial tests). As during free whisking, the most strongly modulated cells 98 were the cells that decreased their firing rate (Fig. 1h,j bottom, both P < 0.05, Mann-Whitney 99 U-test).

100

101 Cellular mechanisms of vibrissa motor cortex suppression

102 Thus, the transition from rest (retracted whiskers, no movement) to whisker behaviors (pro- 103 tracted whiskers, whisking) leads to a decrease of VMC activity. In cortical physiology, this is 104 a highly unusual result. In the somatosensory system and the visual system, relevant stimuli 105 lead to an increase in population activity24. To explore the cellular basis of the decrease of 106 VMC activity during whisking, we habituated rats to head-fixation and performed juxtacellu- 107 lar recording, nanostimulation and whole-cell recordings from VMC putative layer 5 neurons,

94 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 5

108 the output layer1. We focused on social touch, an engaging stimulus18, which strongly acti- 109 vates primary somatosensory cortex (S1)19,20 and medial prefrontal cortex25. During record- 110 ings, we staged facial interactions of the head-fixed rats with stimulus rats in front of them 111 (Fig. 2a). In agreement with the recordings in freely moving rats, we found that VMC activity 112 strongly decreased during social facial touch episodes. As shown in Fig. 2b the rat protracted 113 the whiskers during nose-to-nose touch and a juxtacellularly recorded neuron discharged few- 114 er APs than at baseline. Across the population of neurons, we found a significant decrease of 115 spiking during social facial touch compared to baseline (Fig 2c, median 2.5/2.0 Hz, base- 116 line/social touch, P = 0.0079, N = 21 cells, Wilcoxon signed-rank test). To investigate if the 117 decrease in spiking was due to a decrease in cell excitability, we evoked APs in single layer 5 118 neurons at baseline and during social touch episodes, using a nanostimulation protocol (Fig. 119 2d, see Methods). Across the population, we found that layer 5 neurons were indeed much 120 less excitable during social facial touch than during baseline (Fig. 2e, median evoked rate 121 14.6/4.8 Hz, baseline/social touch, P = 0.0125, N = 15, Wilcoxon signed-rank test). To inves- 122 tigate the underlying intracellular signals, we targeted whole-cell patch-clamp recordings to 123 the deep layers of VMC during social facial touch (Fig. 2f). In agreement with the reduced 124 excitability, we found that the neurons were slightly but significantly more hyperpolarized 125 during social facial touch than at baseline, on average by 1.5 mV (Fig. 1f, P = 0.0171, N = 10 126 cells, paired t-test). Some cells showed a reduction in the membrane potential coefficient of 127 variation during social touch (e.g. Fig. 1e), but across the population, there was no significant 128 change (P > 0.05, N = 10 cells, paired t-test). Both the dampening of spiking evoked by juxta- 129 somal nanostimulation and the observed hyperpolarization point to increased somatic inhibi- 130 tion in VMC during whisker movement.

131

132 Whisker protraction and social touch drive suppression of vibrissa motor cortex

133 Even though social facial touch is generally associated with whisker movement and whisker 134 protraction (Fig 1d), there is a large variability in the whisking between touch episodes18,19,26. 135 We decided to exploit this fact to disentangle whether the suppression of VMC activity during 136 social facial touch (Fig. 2b) was due to the nose-to-nose touch, the coincidental whisker pro- 137 traction and increased whisking amplitude, or perhaps a combination thereof. To this end, we 138 juxtacellularly recorded an additional set of cells during social facial touch episodes, where 139 we now also simultaneously recorded and tracked the whisker angle of the contralateral 140 whiskers using high-speed videography, using a robust method which captures most aspects 141 of the whisker movements (see Methods). We then used likelihood maximization to fit a Pois- 142 son model (with 1-ms bins) to the spike trains27, where the instantaneous firing rate depends 143 linearly on nose-to-nose touch (a binary variable), the whisker angle and the whisking ampli- 144 tude by the coefficients βNose, βAngle and βAmpl (Fig. 3a-b, see Methods). After recording, we 145 labeled and recovered cells, both Ctip2-positive cells (putative thick-tufted pyramidal tract 146 (PT-type) neurons28, Fig 3a-b) and Ctip2-negative cells (putative thin-tufted intratelencephalic 147 (IT-type) neurons, example Fig. S3d-e). Across all cells, we found that the cells were sup- 148 pressed by both nose touch (Fig. 3c, median βNose = − 0.60, P = 0.0067, N = 32 cells, Wilcox- -1 149 on signed-rank test) and by whisker protraction (Fig. 3c, median βAngle = − 0.026 (°) , P = 150 0.020, N = 32 cells, Wilcoxon signed-rank test). We did not find a systematic dependence on -1 151 the whisking amplitude across the population (median βAmpl = − 0.026 (°) , P > 0.05, N = 32 152 cells, Wilcoxon signed-rank test). When we used a likelihood ratio test to select single cells, 153 which were significantly modulated by amplitude (at P < 0.05, see Methods), the results were

95 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 6

154 also mixed (Fig S3b): 10 cells were significantly suppressed, 6 cells were significantly acti- 155 vated, and 16 cells were not significantly modulated. However, we note suppressed cells were 156 more strongly modulated that the activated cells: (median 0.221/0.128 for sup-

157 pressed/activated cells, P = 0.00025, Mann-Whitney U-test). In !"#$our subset of labeled cells, we 158 did not find indications, that PT or IT type cells had different𝛽𝛽 response= patterns; both cell 159 types were generally suppressed by nose touch and whisker protraction (data not shown). To 160 control for possible collinearity, we also fitted a model, where we performed stepwise orthog- 161 onalization of the predictor vectors ‘nose touch’, ‘angle’ and ‘amplitude’ using the Gram- 162 Schmidt algorithm. In this case, we found the same pattern: the ’s and the ’s were 163 significantly negative (Fig. S3f) 𝛽𝛽!"#$ 𝛽𝛽!"#$% 164 Since each cell is associated with an individual estimate of the baseline firing rate and de- 165 pendence on nose touch and whisker protraction, we could evaluate our model for all cells to 166 estimate the population activity during various whisker behaviors. We first compared the 167 baseline firing rate ) to the firing rate during nose touch 168 ). In agreement with Fig. 2c, we found that the population activity is suppressed during (𝜆𝜆!"#$ = exp (𝛽𝛽0) (𝜆𝜆𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡ℎ = exp (𝛽𝛽0 + 169 nose touch (Fig dc, median 1.69/1.15 Hz, baseline/nose touch, P = 0.036, N = 32 cells, paired 𝛽𝛽𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁) 170 t-test), even in the absence of whisker protraction. When we calculated the modulation index 171 resulting from comparing rest to nose touch (Fig 3e, left, median index: −0.29, P = 0.0067, N 172 = 32 cells, Wilcoxon signed-rank test), comparing rest to 15° whisker protraction (Fig 3e, 173 middle, median index: −0.19, P = 0.020, N = 32 cells, Wilcoxon signed-rank test) and from 174 comparing rest to nose touch coinciding with 15° whisker protraction (Fig 3e, right, median 175 index: −0.40, P = 0.00041, N = 32 cells, Wilcoxon signed-rank test), we found that all condi- 176 tions lead to a suppression of the population activity. We conclude, that the suppression of 177 activity during exploratory whisking in air and during social facial touch, which we observe in 178 behaving animals (Fig 1 & 2), likely results from a suppression due to both nose touch and 179 coincidental whisker protraction (Fig 3c).

180

181 Activation of vibrissa motor cortex by microstimulation

182 In order to better understand the role of VMC activity in control of contralateral movement 183 during ongoing whisking, we decided to test the effect of increasing VMC activity by intra- 184 cortical microstimulation at a behavioral time scale in the range where we observed modula- 185 tion of VMC activity23. To this end, we unilaterally microstimulated VMC deep layer cells 186 during bouts of free whisking using 1 s long stimulation pulse trains randomly preceded or 187 followed by a 1 s long break (“sham stimulation”) (Fig. 4a, see Methods). The dominant ef- 188 fect of such stimulation was the retraction of the contralateral whiskers (Fig. 4b, 1.2 ± 0.7°/s 189 vs. –2.4 ± 1.5°/s, sham vs. stimulation, P = 0.0000245, paired t-test). We also observed a 190 small increase in the contralateral whisking power during the stimulation pulses, suggesting 191 that we induced extra movement of the contralateral whiskers (11% increase, Fig. 4c, 194 ± 192 64(°)2 vs. 216 ± 64(°)2, P = 0.0151, paired t-test). We wondered, if the extra induced back- 193 wards movement might be enough to influence social touch behavior, so we also performed a 194 set of experiments, where we unilaterally microstimulated VMC deep layers during social 195 facial touch episodes. When comparing the duration of social facial touch episodes from first 196 to last whisker touch, we found that microstimulation significantly shortens the whisker 197 touches (Figure S4, mean duration 0.692/0.973 s, stimulation/sham, P = 0.0000011, LME

96 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 7

198 model), consistent with the finding that VMC activation induces whisker retraction to abort 199 social facial touch.

200

97 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 8

201 Blockade of vibrissa motor cortex

202 We wondered, how whisking would be affected by VMC inactivation and therefore pharma- 203 cologically blocked VMC deep layer activity unilaterally by injection of lidocaine. As shown 204 in an example experiment prior to lidocaine injection, the rat’s whiskers were positioned 205 symmetrically (Fig. 5a left). After lidocaine injection, the whiskers were asymmetric and 206 more protracted contralaterally (Fig. 5a right). Similarly, the rat whisked with equal whisking 207 amplitude ipsilaterally and contralaterally prior to lidocaine injection (Fig. 5b, left), but 208 whisked with much larger amplitude on the contralateral side than on the ipsilateral side after 209 blockade (Fig. 5b right). The same observations were made across a series of experiments. 210 Injection of lidocaine solution (Fig. 5c left, P = 0.0052, N = 10, paired t-test) but not of ringer 211 solution (Fig. 5c right) led to a significant increase in the contralateral whisking power. Simi- 212 larly, injection of lidocaine but not of ringer solution led to a protraction of the contralateral 213 whiskers by an average of 21° (Fig. 5d, –7.0 ± 16.3° vs. 14.2 ± 7.3°, P = 0.0012, N = 10, 214 paired t-test). Neither injection of ringer nor lidocaine had an effect on the set angle of the 215 ipsilateral whiskers. When we unilaterally blocked excitatory currents in the VMC by super- 216 fusion of APV (an NMDA antagonist) and NBQX (an AMPA antagonist) in lightly anaesthe- 217 tized rats (Figure S5, N = 3 rats), and when we blocked VMC activity by injection of musci- 218 mol (a GABAA agonist) in lightly anaesthetized mice (Figure S6, N = 4 mice), we saw the 219 same effects: protraction of contralateral whiskers and increased contralateral whisker move- 220 ments.

98 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 9

221 DISCUSSION:

222 Summary

223 Most work on the mammalian motor cortex has focused on a role of this cortical area in 224 movement generation11,12. It is therefore surprising that our observations coherently indicate 225 that a prime function of VMC activity might be to suppress behavior: (i) When the rat engag- 226 es in whisker-related behavior (protracted whiskers, whisker movements), we see a decrease 227 in spiking activity in the VMC output layers (Fig. 1-3). (ii) VMC microstimulation leads to 228 retractive movements, as if to abort behavior (Fig. 4). (iii) VMC blockade disinhibits contrala- 229 teral whisker movements and leads to contralateral whisker protraction, as if to engage in be- 230 havior (Fig. 5). Our observations are difficult to reconcile with the classic model, where the 231 prime role of VMC activity is whisker protraction and the generation of movement8. Instead 232 the data support a model where VMC activity suppresses whisker behavior, perhaps by gating 233 a downstream whisking central pattern generator7,15,29,30.

234

235 Relation to previous vibrissa motor cortex studies

236 The whisker motor plant31,32 and vibrissa motor neurons33 are laid out for the fine control of 237 individual whisker protraction. This is in line with a prime function of vibrissal touch in pal- 238 pation of objects, obstacles and conspecifics in front of the animal21,22. In light of the speciali- 239 zation of the motor plant and motor neurons for whisker protraction, our observation that 240 VMC deep layer microstimulation leads to whisker retraction is quite surprising. This retrac- 241 tion result is in agreement with previous studies, which have all reported that VMC mi- 242 crostimulation2–6, single-cell stimulation7 and optogenetic stimulation8 elicit with few excep- 243 tions5,8 whisker retraction. The retraction movements let it appear unlikely that an increase in 244 VMC activity drives vibrissal touch (which is associated with whisker protraction, Fig. 1d). 245 Rather, it suggests, that the role of VMC is to abort undesired whisking behavior. This idea is 246 also supported by the unexpected increase of contralateral whisking following acute VMC 247 blockade. Our observation that a reduction in motor cortex activity increases movement is 248 consistent with observations on whisking patterns after VMC lesions: Whisking persists after 249 VMC ablation21 and blockade8,34, VMC ablation spares large-amplitude whisking, but reduces 250 small whisker movements9, and unilateral VMC lesions increases contralateral whisking pow- 251 er10.

252 While the bulk of our data point to motor suppressive effects of VMC activity, some of our 253 results also point to a role of VMC cells in movement generation. Thus, a small subset of 254 VMC cells weakly increased their firing rate during movements (Fig 1-2). Further, general- 255 ized linear modeling of VMC activity revealed, that the relationship between VMC activity 256 and whisking amplitude was mixed with no obvious pattern (Fig 3). This is consistent with 257 previous studies, which have reported both negative and positive correlations between activity 258 of single VMC cells and whisking power14–17. A novelty of our study is that we used statisti- 259 cal modeling to analyze all spikes and relate them to naturalistic behavior (whisker angle, 260 amplitude and social touch episodes), i.e. we did not only analyze whisking in air14–17 and we 261 did not exclude periods, where the amplitude was low15 (breaks in whisking are also a part of 262 natural whisking patterns). Speaking more generally, movement suppression and movement 263 generation are probably inseparable aspects of motor control.

264

99 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 10

265 Is the functional output of vibrissa motor cortex a decrease in spikes to downstream targets?

266 The output of vibrissa motor cortex is thought to play two major roles: First, it is supposed to 267 control whisker movement, presumably by gating a downstream whisking central pattern gen- 268 erator in the brainstem (PT-type neurons)15,29,30. Secondly, it is supposed to transmit an effer- 269 ent, internal signal to sensory cortices, so they can disentangle afferent sensory signals due to 270 touch stimulation of the whiskers from sensory signals due to self-generated whisker move- 271 ment (IT-type neurons)22,35,36. Previous investigations of the relationship between VMC activ- 272 ity and whisking have found that the overall modulation of VMC activity due to whisking is 273 weak, although single cells exist which correlate with the whisking amplitude14–17 and the 274 whisker angle15. These single cells have previously been found in roughly equal proportions, 275 and it has remained unclear what the prime ‘functional’ output of the VMC might be15,17.

276 In neurophysiology we normally observe that cells respond to behavior with an increase of 277 activity. This is the case in the somatosensory system37 and the visual system38, and has been 278 proposed to be a governing principle for cortical information processing24. Thus, our observa- 279 tion that the ‘functional’ response of VMC cells during various whisker behaviors is a de- 280 crease in spiking is highly surprising. The whisker motor plant is laid out for controlled for- 281 ward movement, yet we found that VMC activity decreases with whisker protraction (Fig. 282 1,3). Social facial touch is a very engaging stimulus, where correct sensorimotor computation 283 is of high ecological importance18, yet we robustly see a decrease in VMC activity and a de- 284 crease in VMC excitability during social touch episodes (Fig 1-3). For comparison, previous 285 studies have found that social facial touch very strongly activates primary somatosensory cor- 286 tex (S1)19,20 and medial prefrontal cortex25. Cells, where the functional response is a hyperpo- 287 larization, are rare but not unknown (e.g. the photoreceptor of the mammalian eye39), yet our 288 findings are very unusual for a primary cortical area.

289

290 Motor suppressive effects of motor cortex

291 Motor effects of motor cortex lesions in rats are subtle, as many simple behaviors (e.g. loco- 292 motion) persist after decortication40. Motor cortex lesions are associated with performance 293 deficits in several movement related tasks, but at least some of these deficits are not primarily 294 due to deficits in the generation of movement, but to deficits in the control and suppression of 295 movement40. For instance, rats can perform long sequences of skilled, learned motor behav- 296 iors after motor cortex ablation, but motor cortex is required for them to learn a task of behav- 297 ioral inhibition (they must learn to postpone lever presses)41. When swimming, intact rats hold 298 their forelimbs still and swim with only their hindlimbs. After forelimb motor cortex lesions, 299 however, rats swim with their forelimbs also42. After learning a go/no-go whisker task, where 300 mice must lick to receive a water reward, motor cortex inactivation does not significantly de- 301 crease the licking at correct times, but massively increases the ‘false alarm’ licking rate 302 (where licking should be suppressed)34,43. Human patients with frontal lesions are notorious 303 for their lack of behavioral control and do things, they should not do40, rodents with lesions in 304 motor cortex often perform movements, that should rather be suppressed.

305 Motor-suppressive effects are impossible to detect in classic motor mapping experiments in 306 lightly anaesthetized animals1–7,44, but human patients report an inability to move as a promi- 307 nent effect of intra-operative stimulation of motor cortex45–50. The existence of these negative 308 motor areas (NMAs) in M1 where stimulation elicit an inhibition of movement is a robust

100 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 11

309 result, but even in humans there has historically been a strong bias towards the study of posi- 310 tive motor effects of M1 stimulation, and consequently the function of NMAs in the inhibito- 311 ry control of movement is still poorly understood46. Furthermore, a complication is that NMA 312 stimulation can elicit positive movements with an increasing stimulation current50, and may as 313 such simply be missed, if only positive stimulation effects are evaluated46.

314

315 Is rat vibrissa motor cortex different from primate motor cortex?

316 Our observation that a decrease in VMC activity leads to whisker protraction is incompatible 317 with a model where VMC PT-type neurons synapse directly onto whisker motor neurons in 318 the facial nucleus. This direct wiring pattern from motor cortex to motor neurons is famously 319 present in primate hand motor cortex51, where overwhelming evidence suggests, that in con- 320 trast to our observations of rat VMC, activity mainly correlates positively with movement11,52. 321 Indeed, there is only a very sparse direct projection from VMC to the whisker motor 322 neurons53 with the vast majority of VMC PT-type neurons targeting brainstem 323 interneurons54,55.

324 It is worth noting, that even in primates, the predominant wiring pattern of corticobulbar and 325 corticospinal projections from M1 is to brainstem and spinal interneurons51. The monosynap- 326 tic projections from M1 to motor neurons innervating distal limb muscles is an exception, 327 which evolved in primates in parallel with the evolution of skilled digit movements51. Spike- 328 triggered averaging techniques used in monkeys have all showed that M1 neuron spikes can 329 predict both EMG peaks and troughs, which suggest that M1 neurons commonly have sup- 330 pressive effects on motorneuronal pools56–58. Recent single-cell recordings in monkeys have 331 shown, that also in primates, some M1 cells correlate negatively with movement: Both premo- 332 tor neurons59 and indeed M160 and M1 pyramidal tract neurons61,62 respond with mirror neu- 333 ron activity to the observation of actions, even when the monkey is not moving, a kind of 334 “monkey see, monkey not do” response63. Similarly, although some muscle weakness is a 335 symptom in patients with M1 or pyramidal tract lesions, the prominent symptoms are ataxia 336 (loss of control over movements), spasticity, clonus, hyperexcitability of reflexes64. Spastic 337 paralysis can be managed by high doses of muscle relaxants to reduce the output from the 338 spinal cord or by sectioning the dorsal roots, suggesting that it represents an abnormal in- 339 crease in muscular input from the spinal cord due to a net loss of descending inhibition from 340 M164.

341

342 Conclusion

343 Action suppression is vital for behavior and numerous studies point to a frontal cortical loca- 344 tion of this important cognitive capacity40. Our observations suggest that the classic work on 345 the role of motor cortex in movement generation should be complemented by a more exten- 346 sive investigation of motor suppressive functions of motor cortices.

347

101 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 12

348 REFERENCES

349 1. Neafsey, E. J. et al. The organization of the rat motor cortex: A microstimulation 350 mapping study. Brain Res. Rev. 11, 77–96 (1986). 351 2. Brecht, M. et al. Organization of rat vibrissa motor cortex and adjacent areas according 352 to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells. 353 J. Comp. Neurol. 479, 360–373 (2004). 354 3. Hall, R. D. & Lindholm, E. P. Organization of motor and somatosensory neocortex in 355 the albino rat. Brain Res. 66, 23–38 (1974). 356 4. Gioanni, Y. & Lamarche, M. A reappraisal of rat motor cortex organization by 357 intracortical microstimulation. Brain Res. 344, 49–61 (1985). 358 5. Haiss, F. & Schwarz, C. Spatial segregation of different modes of movement control in 359 the whisker representation of rat primary motor cortex. J. Neurosci. 25, 1579–1587 360 (2005). 361 6. Tandon, S., Kambi, N. & Jain, N. Overlapping representations of the neck and 362 whiskers in the rat motor cortex revealed by mapping at different anaesthetic depths. 363 Eur. J. Neurosci. 27, 228–237 (2008). 364 7. Brecht, M., Schneider, M., Sakmann, B. & Margrie, T. Whisker movements evoked by 365 stimulation of single pyramidal cells in rat motor cortex. 427, (2004). 366 8. Matyas, F. et al. Motor control by sensory cortex. Science 330, 1240–1243 (2010). 367 9. Semba, K. & Komisaruk, B. R. Neural substrates of two different rhythmical vibrissal 368 movements in the rat. Neuroscience 12, 761–774 (1984). 369 10. Gao, P., Hattox, A. M., Jones, L. M., Keller, A. & Zeigler, H. P. Whisker motor cortex 370 ablation and whisker movement patterns. Somatosens. Mot. Res. 20, 191–198 (2003). 371 11. Georgopoulos, A. P., Schwartz, A. B. & Kettner, R. E. Neuronal population coding of 372 movement direction. Science (80-. ). 233, 1416–1419 (1986). 373 12. Lemon, R. The output map of the primate motor cortex. Trends Neurosci. 11, 501–506 374 (1988). 375 13. Asanuma, H. Recent developments in the study of the columnar arrangement of 376 neurons within the motor cortex. Physiol. Rev. 55, 143–156 (1975). 377 14. Carvell, G. E., Miller, S. A. & Simons, D. J. The relationship of vibrissal motor cortex 378 unit activity to whisking in the awake rat. Somatosens. Mot. Res. 13, 115–127 (1996). 379 15. Hill, D. N., Curtis, J. C., Moore, J. D. & Kleinfeld, D. Primary motor cortex reports 380 efferent control of vibrissa motion on multiple timescales. Neuron 72, 344–356 (2011). 381 16. Friedman, W. A., Zeigler, H. P. & Keller, A. Vibrissae motor cortex unit activity 382 during whisking. J. Neurophysiol. 107, 551–563 (2012). 383 17. Gerdjikov, T. V, Haiss, F., Rodriguez-Sierra, O. E. & Schwarz, C. Rhythmic whisking 384 area (RW) in rat primary motor cortex: an internal monitor of movement-related 385 signals? J. Neurosci. 33, 14193–204 (2013). 386 18. Wolfe, J., Mende, C. & Brecht, M. Social facial touch in rats. Behav. Neurosci. 125, 387 900–910 (2011). 388 19. Bobrov, E., Wolfe, J., Rao, R. P. & Brecht, M. The representation of social facial touch 389 in rat barrel cortex. Curr. Biol. 24, 109–115 (2014). 390 20. Lenschow, C. & Brecht, M. Barrel Cortex Membrane Potential Dynamics in Social 391 Touch. Neuron 85, 718–725 (2015).

102 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 13

392 21. Welker, W. I. Analysis of Sniffing of the Albino Rat. Behaviour 22, 223–244 (1964). 393 22. Kleinfeld, D., Ahissar, E. & Diamond, M. E. Active sensation: insights from the rodent 394 vibrissa sensorimotor system. Curr. Opin. Neurobiol. 16, 435–444 (2006). 395 23. Graziano, M. S. a, Taylor, C. S. R., Moore, T. & Cooke, D. F. The cortical control of 396 movement revisited. Neuron 36, 349–362 (2002). 397 24. Mountcastle, V. An organizing principle for cerebral function: the unit model and the 398 distributed system. The Mindful Brain (1978). 399 25. Lee, E. et al. Enhanced Neuronal Activity in the Medial Prefrontal Cortex during 400 Social Approach Behavior. 36, 6926–6936 (2016). 401 26. Lenschow, C. & Brecht, M. Membrane potential dynamics of barrel cortex neurons 402 during staged episodes of active social touch. 45 (2012). 403 27. MacDonald, C. J., Lepage, K. Q., Eden, U. T. & Eichenbaum, H. Hippocampal ‘time 404 cells’ bridge the gap in memory for discontiguous events. Neuron 71, 737–749 (2011). 405 28. Schiemann, J. et al. Cellular Mechanisms Underlying Behavioral State-Dependent 406 Bidirectional Modulation of Motor Cortex Output. Cell Rep. (2015). 407 doi:10.1016/j.celrep.2015.04.042 408 29. Moore, J. D. et al. Hierarchy of orofacial rhythms revealed through whisking and 409 breathing. Nature 497, 205–210 (2013). 410 30. Deschênes, M. et al. Inhibition, Not Excitation, Drives Rhythmic Whisking. Neuron 411 90, 374–387 (2016). 412 31. Dörfl, J. The musculature of the mystacial vibrissae of the white mouse. J. Anat. 135, 413 147–154 (1982). 414 32. Haidarliu, S., Simony, E., Golomb, D. & Ahissar, E. Muscle architecture in the 415 mystacial pad of the rat. Anat. Rec. 293, 1192–1206 (2010). 416 33. Herfst, L. J. & Brecht, M. Whisker movements evoked by stimulation of single motor 417 neurons in the facial nucleus of the rat. J. Neurophysiol. 99, 2821–32 (2008). 418 34. Huber, D. et al. Multiple dynamic representations in the motor cortex during 419 sensorimotor learning. Nature 484, 473–478 (2012). 420 35. Diamond, M. E. & Arabzadeh, E. Whisker sensory system - From receptor to decision. 421 Prog. Neurobiol. 103, 28–40 (2013). 422 36. Kleinfeld, D. & Deschênes, M. Neuronal basis for object location in the vibrissa 423 scanning sensorimotor system. Neuron 72, 455–468 (2011). 424 37. Talbot, W. H., Darian-Smith, I., Kornhuber, H. H. & Mountcastle, V. B. The sense of 425 flutter-vibration: comparison of the human capacity with response patterns of 426 mechanoreceptive afferents from the monkey hand. J. Neurophysiol. 31, 301–334 427 (1968). 428 38. HUBEL, D. H. & WIESEL, T. N. Receptive fields of single neurones in the cat’s 429 striate cortex. J. Physiol. 148, 574–91 (1959). 430 39. Cao, L.-H., Luo, D.-G. & Yau, K.-W. Light responses of primate and other mammalian 431 cones. Proc. Natl. Acad. Sci. U. S. A. 111, 2752–7 (2014). 432 40. Kolb, B. Functions of the frontal cortex of the rat: a comparative review. Brain Res. 433 320, 65–98 (1984). 434 41. Kawai, R. et al. Motor Cortex Is Required for Learning but Not for Executing a Motor 435 Skill. Neuron 800–812 (2015). doi:10.1016/j.neuron.2015.03.024

103 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 14

436 42. Stoltz, S., Humm, J. L. & Schallert, T. Cortical injury impairs contralateral forelimb 437 immobility during swimming: a simple test for loss of inhibitory motor control. Behav. 438 Brain Res. 106, 127–32 (1999). 439 43. Zagha, E., Ge, X. & McCormick, D. A. Competing Neural Ensembles in Motor Cortex 440 Gate Goal-Directed Motor Output. Neuron 88, 565–577 (2015). 441 44. Berg, R. W. & Kleinfeld, D. Vibrissa movement elicited by rhythmic electrical 442 microstimulation to motor cortex in the aroused rat mimics exploratory whisking. J. 443 Neurophysiol. 90, 2950–2963 (2003). 444 45. Penfield, W. & Rasmussen, T. The Cerebral Cortex of Man. (The Macmillan 445 Company, 1952). doi:10.1212/WNL.1.1.96 446 46. Filevich, E., Kühn, S. & Haggard, P. Negative motor phenomena in cortical 447 stimulation: Implications for inhibitory control of human action. Cortex 48, 1251–1261 448 (2012). 449 47. Ikeda, A. et al. Movement-related potentials associated with bilateral simultaneous and 450 unilateral movements recorded from human supplementary motor area. 451 Electroencephalogr. Clin. Neurophysiol. 95, 323–34 (1995). 452 48. Lüders, H. O., Dinner, D. S., Morris, H. H., Wyllie, E. & Comair, Y. G. Cortical 453 electrical stimulation in humans. The negative motor areas. Adv. Neurol. 67, 115–29 454 (1995). 455 49. Nii, Y., Uematsu, S., Lesser, R. P. & Gordon, B. Does the central sulcus divide motor 456 and sensory functions? Cortical mapping of human hand areas as revealed by electrical 457 stimulation through subdural grid electrodes. Neurology 46, 360–7 (1996). 458 50. Mikuni, N. et al. Evidence for a wide distribution of negative motor areas in the 459 perirolandic cortex. Clin. Neurophysiol. 117, 33–40 (2006). 460 51. Kuypers, H. G. J. M. A New Look at the Organization of the Motor System. Prog. 461 Brain Res. 57, 381–403 (1982). 462 52. Lemon, R. N., Mantel, G. W. & Muir, R. B. Corticospinal facilitation of hand muscles 463 during voluntary movement in the conscious monkey. J. Physiol. 381, 497–527 (1986). 464 53. Grinevich, V., Brecht, M. & Osten, P. Monosynaptic pathway from rat vibrissa motor 465 cortex to facial motor neurons revealed by lentivirus-based axonal tracing. J. Neurosci. 466 25, 8250–8258 (2005). 467 54. Sreenivasan, V., Karmakar, K., Rijli, F. M. & Petersen, C. C. H. Parallel pathways 468 from motor and somatosensory cortex for controlling whisker movements in mice. Eur. 469 J. Neurosci. 41, 354–367 (2015). 470 55. Takatoh, J. et al. New modules are added to vibrissal premotor circuitry with the 471 emergence of exploratory whisking. Neuron 77, 346–60 (2013). 472 56. Buys, E. J., Lemon, R. N., Mantel, G. W. & Muir, R. B. Selective facilitation of 473 different hand muscles by single corticospinal neurones in the conscious monkey. J. 474 Physiol. 381, 529–49 (1986). 475 57. Cheney, P. D. & Fetz, E. E. Functional classes of primate corticomotoneuronal cells 476 and their relation to active force. J. Neurophysiol. 44, 773–791 (1980). 477 58. Davidson, A. G., Chan, V., O’Dell, R. & Schieber, M. H. Rapid changes in throughput 478 from single motor cortex neurons to muscle activity. Science 318, 1934–7 (2007). 479 59. Bonini, L., Maranesi, M., Livi, A., Fogassi, L. & Rizzolatti, G. Ventral premotor 480 neurons encoding representations of action during self and others’ inaction. Curr. Biol.

104 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 15

481 24, 1611–1614 (2014). 482 60. Dushanova, J. & Donoghue, J. Neurons in primary motor cortex engaged during action 483 observation. Eur. J. Neurosci. 31, 386–398 (2010). 484 61. Vigneswaran, G., Philipp, R., Lemon, R. N. & Kraskov, A. M1 corticospinal mirror 485 neurons and their role in movement suppression during action observation. Curr. Biol. 486 23, 236–243 (2013). 487 62. Kraskov, A. et al. Corticospinal mirror neurons. Philos. Trans. R. Soc. Lond. B. Biol. 488 Sci. 369, 20130174 (2014). 489 63. Schieber, M. H. Mirror neurons: reflecting on the motor cortex and spinal cord. Curr. 490 Biol. 23, R151–2 (2013). 491 64. Purves, D. Neuroscience Third Edition. Vascular 3, (2004). 492 493

105 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 16

494 Acknowledgements We thank Brigitte Geue, Undine Schneeweiß and Juliane Diederichs for 495 technical assistance and Viktor Bahr and Falk Mielke for assistance with programming. We 496 thank Maria Rüsseler for assistance with video tracking and Rajnish P. Rao and Evgeny 497 Bobrov for sharing tracked whisker traces of behaving rats. We thank Andreea Neukirchner, 498 Edith Chorev, Saikat Ray, Peter Bennett and Ann Clemens for comments on the manuscript. 499 We thank Sara Helgheim Tawfiq for behavior drawings. This work was supported by Hum- 500 boldt-Universität zu Berlin, the Bernstein Center for Computational Neuroscience Berlin, the 501 German Federal Ministry of Education and Research (BMBF, Förderkennzeichen 502 01GQ1001A) and NeuroCure. M.B. was a recipient of a European Research Council grant 503 and the Gottfried Wilhelm Leibniz Prize.

504 Competing interests statement The authors declare that they have no competing financial 505 interests. 506 Correspondence and requests for materials should be addressed to M.B. 507 ([email protected]).

508 Contributions C.L.E, G.D. and C.L. performed experiments and analysis. M.B. supervised 509 the study. All authors contributed to writing the manuscript.

510

106 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 17

The correct label for the x axes in Figure 1e,g,i is “Time (s)”; not 1e,g,i is Figure “Time in label for the x axes The correct Correction: 1j is 0.0018; not 0.00018. in Figure P value Also, the correct (Hz).” “Time

511 FIGURE 1: Decrease of vibrissa motor cortex activity during vibrissal touch

512 (a) The vibrissa motor cortex (“VMC”, light blue color) is a large frontal area. Soma- 513 tosensory (“S1”, white) and motor (colored) ratunculus shown above. 514 (b) Experimental setup (‘Social gap paradigm’, Wolfe et al. 2011) for recording VMC 515 activity during social facial interactions in freely moving rats: A stimulus rat and a 516 rat with implanted tetrodes for recording are placed on two platforms (25 x 30 cm), 517 separated by a gap (~20 cm, varied slightly according to the individual size of the 518 rats). All experiments were performed in darkness, under infrared illumination. 519 (c) Sketches of four whisking patterns: rest (whiskers not moving), free whisking (self- 520 initiated exploratory whisking in air), object touch (whisking onto objects) and social 521 touch (social touch of a conspecific). 522 (d) Top: Comparison of whisking set angle during rest to free whisking, object touch and 523 social touch (All P < 0.001, t-tests, plot shows mean ± SD). Bottom: Same for whisk- 524 ing power (plot shows median ± 25% and 75% quartiles). 525 (e) PSTH of activity of a layer 5 VMC neuron aligned to the onset of free whisking. A 526 significant firing rate-decrease is observed.

107 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 18

527 (f) Top: Scatterplot of the firing rate during rest vs. free whisking for VMC layer 5 cells. 528 The population activity is lower during free whisking (P = 0.00010, Wilcoxon 529 signed-rank test, dark grey line indicates slope). Cells with significantly decreased 530 activity during free whisking (red dots), cells with significantly increased activity 531 during free whisking (blue dots), cells not significantly modulated (gray dots). Bot- 532 tom: Modulation index of significantly rate-increasing and rate-decreasing cells. 533 Rate-decreasing cells are more strongly modulated (P = 0.0148, Mann-Whitney U- 534 test). 535 (g,h) Same as (e,f) for rest vs. object touch. 536 (i,j) Same as (e,f) for rest vs. social touch. 537

108 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 19

538

539 540 FIGURE 2: Decreased activity, decreased excitability and hyperpolarization of vibrissa 541 motor cortex during social touch

542 (a) VMC recording and nanostimulation in head-fixed rats during staged social touch. 543 (b) Top: Example juxtacellular recording in a VMC layer 5 neuron during a social facial 544 interaction, showing a reduction in APs during social touch (nose-to-nose touch indi- 545 cated by grey bar). Bottom: Angle of contralateral whisker (C2, protraction plotted 546 upwards).

109 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 20

547 (c) Scatterplot of firing rate of VMC layer 5 cells during social facial touch and baseline 548 (P = 0.0079, Wilcoxon signed-rank test). 549 (d) Assessment of cell excitability by nanostimulation. Top: Filtered voltage trace of a 550 VMC layer 5 cell. The evoked firing rate during nanostimulation is higher at baseline, 551 than during social touch (indicated by grey bar). Middle: Unfiltered voltage trace. Bot- 552 tom: Nanostimulation current steps. 553 (e) Scatterplot of the firing rate of VMC layer 5 cells when stimulated during social facial 554 touch and baseline (P = 0.0125, Wilcoxon signed-rank test). 555 (f) Top: Example whole-cell patch clamp recording from a VMC layer 5 cell showing a 556 hyperpolarization of the membrane potential during social facial touch (duration of 557 nose-to-nose touch indicated by grey bar). Middle: Zoom of the above trace (Spikes 558 clipped). Bottom: Angle of contralateral whisker (C2). 559 (g) Scatterplot of the membrane potential (Vm) of VMC layer 5 cells during social facial 560 touch and baseline (P = 0.0171, paired t-test).

561

110 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 21

562 563 FIGURE 3: Vibrissa motor cortex activity is additively suppressed by both nose-to-nose 564 touch and whisker protraction.

565 (a) Example soma of a juxtacellularly labeled neuron in layer 5B of VMC. Top: Overview 566 of coronal section of the VMC showing a wide layer 5, which contains a large fraction 567 of Ctip2-positive, putative thick-tufted pyramidal tract (PT-type) neurons (Red chan- 568 nel = Ctip2, white lines indicate layer boundaries traced on brightfield image, white 569 square indicates location of labeled soma: 1.25mm deep, 1.21 mm medial, uncorrected 570 for shrinkage). Bottom: Close up image of the juxtacellularly recorded soma (labeled 571 by biocytin filling, green channel), which is Ctip2-positive (red channel). 572 (b) Example recorded data and fitted model from the neuron shown in (a). The top traces 573 show the occurrence of nose-no-nose touches (grey bars), the juxtacellular recording 574 trace with spikes (high-pass filtered at 300 Hz, top trace, scale bar = 1 mV) and the 575 whisker angle and whisking amplitude (tracked by high-speed videography, scale bar 576 = 5°). Below we show the estimate of the instantaneous firing rate of the best fitted 577 model (‘MLE model’, green line, smoothed with a Gaussian with σ = 75 ms for plot- 578 ting, real model is run with 1-ms bins) plotted on top of an estimate of the observed 579 firing rate (‘Smoothed SpikeTrain’, grey area, calculated by convolving the spike train 580 with a Gaussian with σ = 75 ms, clipped at 10 Hz for plotting). This cell was sup- 581 pressed by nose touch, whisker protraction and by increased whisking amplitude -1 582 (maximum likelihood estimates: β0 = 0.68, βNose = −1.40, βAngle = −0.06 (°) , βAmpl = 583 −0.20 (°)-1)

111 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 22

584 (c) Across the population, VMC activity is significantly suppressed by nose-no-nose 585 touch and whisker protraction (median βNose < 0, βAmpl < 0, both P < 0.05), but not sig- 586 nificantly modulated by changes in whisking amplitude. 587 (d) Evaluating the MLE model to estimate the firing rate at rest ) and dur- 588 ing nose touch ) recapitulates the finding from Fig 2c: nose- (𝜆𝜆𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 = exp (𝛽𝛽0) 589 to-nose touch suppresses VMC activity. (𝜆𝜆𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡ℎ = exp (𝛽𝛽0 + 𝛽𝛽𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁) 590 (e) Evaluating the MLE model during three behavioral states to demonstrate, that the sup- 591 pression due to nose touch and whisker protraction is additive: nose touch in the ab- 592 sence of whisker protraction (left), 15° whisker protraction in absence of nose touch 593 (middle) and nose touch coinciding with 15° whisker protraction (right) all suppress 594 VMC activity compared to rest.

595

112 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 23

596 597 FIGURE S3: Whisker tracking procedure and additional modeling

598 (a) Example high-speed (250 frames/s) video frame showing whiskers of a head-fixed rat 599 during a juxtacellular recording experiment. The pivot point (red dot) and the whisker 600 tracking ROI (green dots) are manually clicked for tracking each video. 601 (b) Example traces demonstrating the tracking procedure. We rotated adjacent frames 602 around the pivot point shown in (a) to maximize the correlation between the frames 603 within the whisking ROI (‘Pearson’s ρ‘, top trace) and estimated the mean change in 604 angles between adjacent frames (‘ΔAngle’, middle traces). Datapoints with sudden 605 spikes in the correlation between frames due to video artifacts were removed from the 606 traces (example marked by black arrow). To estimate the whisking angle, we linearly 607 interpolated, numerically integrated and band-pass filtered the change in angle be- 608 tween frames (‘Angle’, bottom trace). Grey bar indicates a nose-to-nose touch. 609 (c) Top: Distribution of βAmpl for all cells is not different from zero (P = 0.204, Wilcoxon 610 signed-rank test, also shown in Fig 3c). Bottom: When we plot only significant cells 611 (assessed by a likelihood ratio test), the pattern is mixed: 10 cells are suppressed (red 612 bars) and 6 cells are activated (blue bars). As a population, they are not different from

113 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 24

613 zero (P = 0.098, Wilcoxon signed-rank test), but we note that the suppressed cells tend 614 to be more strongly modulated that the activated cells: (median 0.221/0.128 615 for suppressed/activated cells, P = 0.00025, Mann-Whitney U-test). 𝛽𝛽!"#$ = 616 (d) Soma of example juxtacellularly labeled Ctip2-negative cell. 617 (e) Example recorded data and fitted model from the neuron shown in (d). The top traces 618 show the occurrence of nose-no-nose touches (grey bars), the juxtacellular recording 619 trace with spikes (high-pass filtered at 300 Hz, top trace) and the whisker angle and 620 whisking amplitude (tracked by high-speed videography). Below we show the esti- 621 mate of the instantaneous firing rate of the best fitted model (green line, smoothed 622 with a Gaussian with σ = 75 ms) plotted on top of an estimate of the observed firing 623 rate (grey area, calculated by convolving the spike train with a Gaussian with σ = 75 624 ms, clipped at 10 Hz for plotting). This cell was suppressed by nose touch, whisker 625 protraction and by increased whisking amplitude (maximum likelihood estimates: β0 = -1 -1 626 0.03, βNose = −0.70, βAngle = −0.30 (°) , βAmpl = −0.38 (°) ) 627 (f) Fitted betas, when we run the model shown in Fig 3 on stepwise orthogonalized data. 628 In this model, measures how the spike rate depends on nose touch, 629 measures how the! spike rate depends on ‘the variation in whisker angle, which is or- !"#$ !"#$% 630 thogonal to variations𝛽𝛽 in nose touch’, and measures how the spike rate depends𝛽𝛽′ 631 on ‘the variation in whisking amplitude, which is orthogonal to variations in nose !"#$ 632 touch and variations in whisker angle’. 𝛽𝛽′ 633 634

114 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 25

635 636 FIGURE 4: Unilateral microstimulation of vibrissa motor cortex in awake rats leads to 637 contralateral whisker retraction.

638 (a) Trace of rat whisking during a microstimulation experiment (C2, protraction plotted 639 upwards). Stimulation is delivered in 1 s long pulse trains (“µStim”, slope denoted by 640 orange line) alternating with 1 s pauses (“Sham”, slope denoted by grey line). 641 (b) Comparison of whisker set angle change during periods of sham stimulation (grey 642 dots) and microstimulation (orange dots, vertical lines indicate means) (P = 0.0002, 643 paired t-test). 644 (c) Same as (b) for whisking power (P = 0.0150, paired t-test).

645

115 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 26

646 647 FIGURE S4: Unilateral microstimulation of vibrissa motor cortex shortens social facial 648 touch episodes.

649 (a) Cumulative histograms of the duration of social facial interactions (from first to last 650 whisker-to-whisker touch) on days with VMC microstimulation during interactions 651 (red lines) and days with sham stimulation during interactions (black lines) for one ex- 652 ample rat. 653 (b) Interactions are shorter with VMC microstimulation than during sham stimulation (N 654 = 4 rats, dots indicate median interaction duration, lines indicate rat-specific slope 655 from LME model, colors indicate rats). 656

657

658

659

660

661

662

116 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 27

663 FIGURE 5: Unilateral blockade of vibrissa motor cortex increases contralateral whisker 664 movement and protraction.

665 (a) Left: Rat whisker set angles at 666 rest before unilateral lidocaine injection 667 (green color). Right: whisker set angles 668 after unilateral lidocaine injection (pink 669 color) in deep layers of VMC. Lido- 670 caine injection leads to a protraction of 671 the contralateral whiskers. 672 (b) Ipsilateral and contralateral 673 whisker traces prior to (green color) 674 and after (pink color) lidocaine injec- 675 tion (protraction plotted upwards). Pri- 676 or to injection, whisking is similar on 677 both sides, after injection the contrala- 678 teral whiskers move more. 679 (c) Left: bilaterally symmetric 680 whisking during baseline (contralateral 681 and ipsilateral whisking power ratio ≈1, 682 green dots) changes to a predominance 683 of contralateral whisking after lido- 684 caine injection into VMC deep layers 685 (pink dots, P = 0.0052, paired t-test, 686 lines indicate means). Right: Control 687 injections of ringer have no such effect, 688 (P > 0.05, paired t-test). 689 (d) Same as (c) for whisking set 690 angle. Lidocaine injection results in 691 contralateral protraction (Ringer: P > 692 0.05, Lidocaine: P = 0.0012, paired t- 693 tests).

694

117 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 28

695 696 FIGURE S5: Unilateral blockade of vibrissa motor cortex (by AMPA & NMDA antago- 697 nists) increases contralateral whisker movement and protraction.

698 (a) Example image of anaesthetized rat after unilateral VMC blockade (right hemisphere) 699 by superfusion of APV (an NMDA antagonist) and NBQX (an AMPA antagonist). 700 (b) Example ipsilateral (blue) and contralateral (red) whisking traces of whisker mi- 701 cromovements, which escape light anaesthesia (Whisker arc 1). The contralateral 702 whiskers are more protracted (~27° vs. ~32°) and the contralateral micromovements 703 have a larger amplitude. 704 (c) After VMC blockade, the whisker set point is higher contralaterally (red markers) than 705 ipsilaterally (blue markers) to the blocked hemisphere (N = 3 rats). 706 (d) After VMC blockade, the whisking power is much higher (~5 fold) in the contralateral 707 whiskers than in the ipsilateral whiskers (Markers indicate ratio of contralateral to ip- 708 silateral whisker power).

709

118 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 29

710 711 FIGURE S6: Unilateral blockade of vibrissa motor cortex (by muscimol injection) in- 712 creases contralateral whisker movement and protraction.

713 (a) Top: Example image of lightly anaesthetized mouse after unilateral VMC blockade 714 (left hemisphere) by muscimol injection, showing protraction of contralateral whisk- 715 ers. Bottom: Example whisking pattern from the same mouse showing large whisker 716 movements contralaterally, and smaller whisker movements ipsilaterally. 717 (b) After VMC blockade, the whisker set point is higher contralaterally (red markers) than 718 ipsilaterally (blue markers) to the blocked hemisphere (N = 4 mice). Round markers 719 indicate that only deep VMC was blocked, square markers indicate that both deep and 720 superficial VMC was blocked. 721 (c) After VMC blockade, the whisking power is much higher (~8 fold) in the contralateral 722 whiskers than in the ipsilateral whiskers (Markers indicate ratio of contralateral to ip- 723 silateral whisker power). Round markers indicate that only deep VMC was blocked, 724 square markers indicate that both deep and superficial VMC was blocked. 725 (d) Example ipsilateral (blue) and contralateral (red) whisking traces of whisker mi- 726 cromovements which escape light anaesthesia (Whisker arc 1) in another mouse, 727 showing the whisking patterns at a longer time scale.

728

729

730

731

119 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 30

732 METHODS:

733 Animal welfare

734 All experimental procedures were performed according to German animal welfare law under 735 the supervision of local ethics committees (Permit no. G0259/09, G0193/14). Wistar rats were 736 purchased from Harlan (Eystrup, Germany). Stimulus animals were housed socially in same- 737 sex cages, and post-surgery implanted animals were housed in single animal cages, but were 738 in visual, olfactory and auditory contact other rats. All animals were kept on a 12h:12h re- 739 versed light/dark cycle with lights off at 8:00 a.m., so that all experiments were performed in 740 the rats’ dark phase. Rats had ad libitum access to food and water.

741 Whisker behavior

742 Behavioral experiments were done using the ”social gap paradigm”18,19,65. The experimental 743 paradigm consists of two elevated platforms, 30 cm long and 25 cm wide surrounded by walls 744 on 3 sides, positioned approximately 20 cm apart. The distance between platforms was varied 745 slightly depending on the size of the rats. The platforms and platform walls were covered with 746 soft black foam mats to provide a dark and nonreflective background and to reduce mechani- 747 cal artifacts in electrophysiological recordings and ultrasound recordings. All experiments 748 were performed in (visual spectrum) darkness or in dim light, and behavior was monitored by 749 monochrome video recording obtained under illumination with infrared light, not visible to 750 the rats. The implanted rat was placed on one platform, and on the other platform we either 751 presented various objects or conspecific rats. The implanted rats were not trained, but would 752 spontaneously engage in investigation of the objects or social facial interactions with conspe- 753 cifics.

754 The rat behavior was recorded at low speed from above with a 25 Hz digital camera, synchro- 755 nized to the electrophysiological data acquisition using TTL pulses to trigger each frame. Ad- 756 ditional 250 Hz high-speed recordings were performed, when the rats were freely whisking 757 over the gap, socially interacting or investigating objects. Typically, recording sessions were 758 performed in four to eight 15 min blocks, where we would present either objects or conspecif- 759 ics (of both sexes) in each block, randomly. The video frames of the 25 Hz videos were la- 760 belled in four categories: ”Free whisking” (Animal freely whisking into air), ”Object touch” 761 (animal touching an object with its nose), ”Social touch” (animal touching a conspecific nose- 762 to-nose) and ”Rest” (animal not whisking). Videos were labeled blind to the spike data.

763 In our assessment of the whisker set angle and whisker power during the various whisker be- 764 haviors, we included a large dataset of already-tracked whisker traces, some of which have 765 previously been published in Bobrov et al. 201419 and Rao et al. 201465.

766 To quantify the whisking behavior, the whisker traces were tracked from the 250 Hz video 767 frames, as previously described18,19,65. We first band-pass filtered the raw tracked whisker 768 trace to remove jitter due to the tracking (2nd order Butterwoth filter from 0.25 to 12.5 Hz). 769 The whisking power was calculated from a spectrogram constructed by performing a Stock- 770 well transform from 0-20 Hz (frequency steps of 0.1 Hz)66, and by integrating the absolute 771 value of the power spectral density in the 0-20 Hz band over time to calculate an average 772 power. The set angle was estimated by calculating the average angle of the whisker trace.

120 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 31

773 Tetrode recording of vibrissa motor cortex activity in freely moving animals

774 In tetrode recording experiments, we used p60 Wistar rats (N = 5, 3 male, 2 female), which 775 were handled for 2-3 days, before being implanted with a tetrode microdrive over the vibrissa 776 motor cortex (centered on 1.5 mm anterior, 1.5 mm lateral from bregma). Before surgery, the 777 rats were briefly anesthetized with isoflurane and then injected i.p. with a dose of 100 mg/kg 778 ketamine and 7.5 mg/kg xylazine. During the surgery, the anesthesia depth was monitored by 779 watching the rat respiration rate, testing the pinch reflex and monitoring whisker micromove- 780 ment. If the rat appeared to be entering a lighter state of anaesthesia, additional alternating 781 doses of 25% of the initial dose in ketamine/xylazine amount or 25% of the corresponding 782 ketamine dose alone were given. Typically, this was needed 1 h after the first injection. Dur- 783 ing the surgery, the rat was placed on a heating pad and was kept at approximately 35°C using 784 a feedback system attached to a rectal temperature probe (Stoelting, Wood Dale, IL, USA). 10 785 mins before the first incision, the scalp was locally anaesthetized by injection of a 1% lido- 786 caine solution. Then the rat was placed in a stereotax, the scalp was cut and the tissue on the 787 skull removed.

788 The implanted microdrive had eight separately movable tetrodes driven by screw microdrives 789 (Harlan 8-drive; Neuralynx, Bozeman, MT, USA). The tetrodes were twisted from 12.5 µm 790 diameter nichrome wire coated with polymide (California Fine Wire Company), cut and ex- 791 amined for quality using light microscopy and gold-plated to a resistance of ca. 300 kOhm in 792 the gold-plating solution using an automatic plating protocol (“nanoZ”, Neuralynx). For tet- 793 rode recordings, a craniotomy of 1x2 mm was made 0.75-2.75 mm anterior and 1-2 mm lat- 794 eral to bregma, corresponding to the coordinates of VMC. Steel screws for stability and two 795 gold screws for grounding the headstage PCB were drilled and inserted into the skull, and the 796 gold screws were soldered and connected to the headstage PCB using silver wire. After fixa- 797 tion of all screws, the dura was removed, the implant fixated in the craniotomy, the cranioto- 798 my sealed with 0.5% agarose and the tetrode drive fixed in place with dental cement (Herae- 799 us). Outer polyimide guiding tubes were arranged in a 2-by-4 grid (d ≈ 500 µm) and contained 800 smaller polyimide tubes, which in turn contained the tetrode wires. Neural signals were rec- 801 orded through a unity-gain headstage preamp and transmitted via a soft tether cable to a digi- 802 tal amplifier and A/D converter (Digital Lynx SX; Neuralynx). The spike signals were ampli- 803 fied by a factor of 10 and then digitized at 32 kHz. The digital signal was bandpass filtered 804 between 600 Hz and 6 kHz. Spike events were detected by crossing of a threshold (typically 805 ~50 µV) and recorded for 1 ms (23 samples - 250 µs before voltage peak and 750 µs after 806 voltage peak). At the end of the experiment, animals were again anaesthetized with a mix of 807 ketamine and xylazine, and the single tetrode tracks were labelled using small electrolytic 808 lesions made by injecting current through the tetrode wire (10 µA for 10 s, tip-negative DC). 809 After lesioning, animals were perfused with phosphate buffer followed by a 4% paraformal- 810 dehyde solution (PFA). Brains were stored overnight in 4% PFA before preparing 150 µm 811 coronal sections. Sections were stained for cytochrome oxidase to reveal the areal and laminar 812 location of tetrode recording sites, which could be calculated from the location of tetrode 813 tracks and lesions. We only analyzed data from recording sites, where the lesion pattern could 814 unanimously identify the tetrode and the recording sites.

815 All spike analysis was done in Matlab (MathWorks, Natick, MA, USA). Spikes were preclus- 816 tered off-line on the basis of their amplitude and principal components by means of a semiau- 817 tomatic clustering algorithm (KlustaKwik by K. D. Harris, Rutgers University). After preclus- 818 tering, the cluster quality was assessed and the clustering refined manually using MClust (A.

121 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 32

819 D. Redish, University of Minnesota). The spike features used for clustering were energy and 820 the first principle component of the waveform. To be included in the analysis as a single unit, 821 clusters had to fulfill the following criteria: first, the L-ratio, a measure of distance between 822 clusters67, was below 0.5. Second, the histogram of inter-spike intervals (ISIs) had to have a 823 shape indicating the presence of a refractory time of 1-2 ms, or have the appearance of a 824 bursty cell (many short ISIs). Flat ISI histograms were indicative of multi-unit activities, and 825 these units were not included. Further requirements were that the firing of the cluster was sta- 826 ble over the course of the recordings and that the cluster did not appear to be ”cut” - that is, so 827 close to noise that many spikes were not detected as spike events.

828 Since we wanted to investigate the contribution of VMC to motor control, we were interested 829 in the spiking activity of the pyramidal projection neurons in layer 51. Due to the different 830 morphology and ion channel populations in the cell membrane, interneurons and pyramidal 831 cells can sometimes be separated based on the shape of the extracellular spike waveform68. 832 We tried various combinations of spike shape parameters such as spike-width, peak-to-trough 833 time, shape of after-hyperpolarization, but none yielded convincingly bimodal distributions 834 which allowed separation of cells into regular-spiking putative pyramids and fast-spiking pu- 835 tative interneurons (data not shown). This may relate to the fact that motor cortex projection 836 neurons have exceedingly narrow spikes69. Since separation by spike shape was not feasible, 837 we instead reduced the number of fast-spiking interneurons by simply excluding very-high 838 firing cells (mean rate during whole recording session above 10 Hz) from the analysis.

839 In the case of all three behaviors, we compared the firing rate during the behavior to the firing 840 rate during ”Rest”, defined as when the rats were not moving the whiskers. To statistically 841 asses, if single cells were significantly modulated by whisker behaviors, we used a bootstrap 842 method: First, we calculated a modulation index: Idx = (Rbehavior –Rbaseline)/(Rbehavior+ Rbaseline) 843 where Rbehavior and Rbaseline is the average firing rate at baseline and during the behavior (e.g. 844 during social touch), respectively. In the bootstrapping, we avoided bias by having a balanced 845 baseline design, where the baseline was defined to be segments of time of equal lengths to the 846 nose-to-nose touches, just prior to the beginning of nose-to-nose touches. We generated a dis- 847 tribution of 10.000 bootstrapped dummy modulation indices by preserving the lengths of the 848 nose-to-nose touches, but randomly placing the start-times within the recordings. If the real 849 modulation index was below the 2.5th or above the 97.5th percentile of the bootstrapped 850 dummy indices (i.e. a two-tailed test at α = 0.05), the cell was taken to be significantly modu- 851 lated.

852 After having assessed that a cell was significantly modulated, we compared significantly 853 modulated cells to ”Rest”. Here we calculated a similar modulation index: Modulation index 854 = (Rbehavior – Rrest)/(Rbehavior+ Rrest), where Rbehavior and Rrest is the average firing rate at ”Rest” 855 and during the behavior of interest (e.g. during social touch), respectively. The response index 856 is symmetric and can take on values between -1 (cell only spikes at ”Rest”) and +1 (cell only 857 spikes during behavior). Thus, cells with negative indices are suppressed during behavior and 858 cells with positive indices are spiking more during behavior. To compare the modulation 859 strength of suppressed and increased cells, we compared the absolute value of this response 860 index.

861 Head-fixed juxtacellular and whole cell patch-clamp recordings in the vibrissa motor 862 cortex

122 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 33

863 In juxtacellular recording experiments, rats were between 35 and 40 days old. Whole cell 864 patch clamp recordings were made in younger animals, aged between P25-P30 at the day of 865 the final experiments. All Wistar rats (male) were handled for 2-3 days, before being implant- 866 ed with a head-fixation post and a recording chamber over the vibrissa motor cortex (1.5 mm 867 anterior, 1.5 mm lateral from bregma). The surgery procedure including anesthesia and prepa- 868 ration of the skull were the same as described above. The head-fixation post and recording 869 chamber were fixed to the skull using a UV-curable adhesive (Kerr) and dental cement 870 (Heraeus). After the first surgery, animals were given 2 days of rest and then habituated to 871 head-fixation over several days. The rat was first head-fixed for 5 minutes in the first head- 872 fixation session, then for an additional 10 minutes with each succeeding session until the rat 873 was comfortable with head-fixation for 60 minutes. During the habituation procedure, the rat 874 was also accustomed to the experimental setup (e.g. microscope light turning on and off, noise 875 from the micromanipulator etc.). Habituation to head-fixation took 2-4 days on average, de- 876 pending on the rat’s behavior. After the habituation procedure a second surgery was per- 877 formed, during which a craniotomy was drilled inside the recording chamber. In the case of 878 patch-clamp recordings the dura was removed using a bent syringe. The preparation was cov- 879 ered with silicone (Kwik-Cast, World Precision Instruments) and additionally protected by a 880 lid closing the cylinder. After the second surgery, the animals recovered for one day before 881 the recording sessions started. 882 883 Juxtacellular and whole-cell patch-clamp recordings were made using glass electrodes made 884 of borosilicate glass tubes (Hilgenberg) pulled to have a resistance of 4 to 6 MΩ. Pipettes 885 were lowered into the cortex with positive pressure (200- 300bar). In both cases, the pipettes 886 was filled with intracellular solution of 135 mM K-gluconate; 10 mM HEPES; 10 mM Na2- 887 phosphocreatine; 4 mM KCl; 4 mM MgATP; and 0.3 mM Na3GTP (pH 7.2). After the pipette 888 reached 150-200µm below the surface the following steps were different depending on the 889 recording type. For juxtacellular recordings a nanostimulation protocol was performed and the 890 pipette was lowered stepwise through the cortex (step size: 3µm) until a cell could be detected 891 by excitability (as previously described70). In the case of patched cells, the positive pressure in 892 the pipette was lowered to 30 bars to search for cells and the pipette was lowered with a step 893 size of 3µm through the cortex. When the pipette resistance increased, suction was applied to 894 establish a gigaohm seal and achieve the whole-cell configuration. The recorded signal was 895 amplified and low-pass filtered at 3 kHz by a patch-clamp amplifier (Dagan) and sampled at 896 25 kHz by a Power1401 data acquisition interface under the control of Spike2 software 897 (CED). All head-fixed recording and stimulation experiments were performed at a depth read- 898 ing of 1423 ± 512 µm (mean ± SD) from the pia, corresponding to putative layer 5 of VMC. 899 Only regular spiking, putative pyramidal neurons were included in the analysis.

900 During head-fixed recording sessions, stimulus rats, hand held by the experimenter, were pre- 901 sented in front of the head-fixed rat the rats were allowed to socially interact. To monitor so- 902 cial interactions, 25 Hz and 250 Hz digital video synchronized to the electrophysiology data 903 was recorded from above by triggering frames and recording from a Spike2 script. The 904 whisker movements were tracked from the 250 Hz videos using a custom written computer 905 software for whisker tracking (Viktor Bahr, adapted from Clack et al. 201271). Behavioral 906 events (beginning and end of nose touches) were labeled in the 25 Hz videos. All video analy- 907 sis was performed blind to the electrophysiology data.

908 To estimate the firing rate change during social facial touch in the head-fixed animals, we 909 computed the average firing rate during 1 second preceding start of social facial touch (begin-

123 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 34

910 ning of nose-to-nose touch), this we used as a baseline firing rate. Firing rates during social 911 facial touch were computed by averaging the firing rate in a 1 s response window after the 912 onset of nose-to-nose touch. In this analysis, we included both juxtacellularly recorded cells, 913 and patched cells which were spiking. To estimate the change in cell excitability, we calculat- 914 ed the average firing rate from juxtacellularly nano-stimulated cells during positive stimula- 915 tion pulses, where the rat was not engaging in social facial touch (the baseline). This we com- 916 pared to the average firing rate during stimulation pulses that happened during a social facial 917 episode (i.e. while the rats were touching nose-to-nose). To estimate the hyperpolarization of 918 the patched cells during social facial touch episodes, we clipped the spikes from the mem- 919 brane potential and compared the average membrane potential during social facial touch epi- 920 sodes to the average membrane potential during 1 second preceding start of social facial touch 921 (the baseline).

922 Automatic whisker tracking in socially interacting, head-fixed animals

923 We used a correlation-based algorithm to automatically track the whisking during juxtacellu- 924 lar recordings in head-fixed rats, all males. We filmed the rats from above using a high-speed 925 camera (250 frames/s, Figure S3a). For each tracked video, we manually clicked a pivot point 926 on the center of the whisker pad and drew a tracking region of interest (ROI) around the 927 whiskers contralateral to the recording craniotomy. To estimate the change in mean whisker 928 angle (‘ΔAngle’) between two adjacent video frames, we calculated the correlation between 929 the two frames within the tracking ROI (Pearson’s ρ calculated between the greyscale values 930 of the pixels) and rotated the previous frame around the pivot point (with nearest-neighbor 931 interpolation), so that the correlation was maximized. This method was very robust, since it 932 considers many whiskers simultaneously, and also worked during nose-to-nose touch epi- 933 sodes, where a few whiskers of the stimulus rat might enter the tracking ROI (Figure S3b, 934 middle). Artifacts from badly tracked video frames were detected as sudden spikes in the cor- 935 relation (Figure S3b, top) and the corresponding estimated values of ΔAngle were removed 936 (by a threshold of ρ > 0.03). To estimate the mean whisker angle (‘Angle’, Figure S3b, bot- 937 tom), we linearly interpolated, numerically integrated and bandpass filtered ΔAngle. For fil- 938 tering, we used a bandpass IIR filter from 5-15 Hz in Matlab, to remove low-frequency drift 939 stemming from the discrete integration. Since our tracking method considers all whiskers 940 within the whisking ROI, our calculated whisker angle should be thought of as the mean devi- 941 ation from the mean set angle of the whiskers, i.e. Angle = 10° corresponds to a mean-field, 942 net 10° degrees whisker protraction, not an absolute whisker angle of 10°.

943 Maximum likelihood modeling

944 We used maximum likelihood modeling to estimate the dependence of VMC activity on the 945 three covariates nose-to-nose touch, whisker angle and whisking amplitude, by fitting a Pois- 946 son model to the spike trains27. First, we binned the spike train in 1 ms bins. We assume, that 947 the discharge of spikes within one time bin is generated by a homogenous Poisson point pro- 948 cess, so that the probability of observing y spikes in a time bin is:

! 𝜆𝜆Δ 𝑝𝑝 𝑦𝑦 𝜆𝜆 = exp(-−1 𝜆𝜆Δ) 949 where ∆ = 1 ms is the width of the time bin and𝑦𝑦! λ > 0 s is the expected discharge rate of the 950 cell. If we assume, that each time bin is independent, the probability of the entire spike train, 951 is:

𝑦𝑦

124 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 35

!! ! 𝜆𝜆 Δ ! 𝑝𝑝 𝑦𝑦 𝜆𝜆 = ! exp(−𝜆𝜆 Δ) 952 where is the observed number of spikes! and𝑦𝑦 ! the expected discharge rate in the i’th time 953 bin, respectively. If we model the expected discharge rate, , so that it depends on some pa- 𝑦𝑦!, 𝜆𝜆! 954 rameters, , we have the log-likelihood function: 𝜆𝜆 𝛽𝛽

ℒ 𝛽𝛽 = log 𝑝𝑝 𝑦𝑦 𝜆𝜆 𝛽𝛽 = 𝑦𝑦! log 𝜆𝜆! + 𝑦𝑦! log Δ − log 𝑦𝑦!! − Δ 𝜆𝜆! 955 For our purpose, we model so that it! depends on the! spike history! and linearly! on a 1-ms 956 interpolated vector indicating nose touch, (either 0 or 1), a vector of the whisker angle, 957 , and a vector of the whisking𝜆𝜆 amplitude, (calculated by quadratically splining 958 the local maxima of the rectified whisker angle).𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 Due to the refractory period of the cell, it is 959 not𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 correct to assume, that all time bins are statistically𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 independent, so following MacDonald 27 960 et al. 2011 , we also include 11 spike history parameters, h1…h11, to model the interspike 961 interval distribution of the cell. The spike history term is binned to 11 successive bins, five 1- 962 ms bins (vectors : no. of spikes in the previous 0-1 ms, 1-2 ms, 2-3 ms, 3-4 ms, 4-5 963 ms,) and six 25-ms bins (vectors : no. of spikes in the previous 5-30 ms, 30-55 ms, ! ! 964 55-80 ms, 80-105𝑛𝑛 ms,… 𝑛𝑛105-130 ms, 130-155 ms). We thus have: 𝑛𝑛! … 𝑛𝑛!!

𝜆𝜆! = exp (𝛽𝛽! + 𝛽𝛽!"#$% ⋅ 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑒𝑒! + 𝛽𝛽!"#$ ⋅ 𝐴𝐴𝐴𝐴𝐴𝐴𝑙𝑙! + 𝛽𝛽!"#$ ⋅ 𝑁𝑁𝑁𝑁𝑁𝑁𝑒𝑒! + ! !! !!" !"# !"!" !"# ℎ! ⋅ 𝑛𝑛!,! + ℎ! ⋅ 𝑛𝑛!,! ) 965 For each cell, we fit the model!! !by adjusting the parameters!!! 966 so that we maximize the log-likelihood function (using ’fminunc’ in Matlab). 𝛽𝛽!, 𝛽𝛽!"#$%, 𝛽𝛽!"#$, 𝛽𝛽!"#$, ℎ! … ℎ!! 967 Since we did not find a dependence of the population activity on whisking amplitude, we also 968 fitted a reduced model to the spike train, which does not depend on the whisking amplitude:

!"#$%"# 𝜆𝜆! = exp (𝛽𝛽! + 𝛽𝛽!"#$% ⋅ 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝑒𝑒! + 𝛽𝛽!"#$ ⋅ 𝑁𝑁𝑁𝑁𝑁𝑁𝑒𝑒! + ! !! !!" !"# !"!" !"# ℎ! ⋅ 𝑛𝑛!,! + ℎ! ⋅ 𝑛𝑛!,! ) 969 We then used a likelihood ratio!!! test between the! !full! model and the reduced model to estimate 970 if single cells are significantly modulated by the whisking amplitude. Since there is one less 971 fitted parameter in the reduced model, the log-likelihood ratio:

𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙ℎ𝑜𝑜𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝐿𝐿𝐿𝐿𝐿𝐿𝑇𝑇 = −2 log 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙ℎ𝑜𝑜𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜 𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 𝑤𝑤𝑤𝑤𝑤𝑤ℎ 𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 !"## !"#$%"# 972 follows a χ2-distribution with one degree= 2(ℒ of freedom− ℒ ( ) ). The p-value of the increase in 973 likelihood due to including whisking amplitude in the model can thus be evaluated using the 974 ‘chi2cdf’ function in Matlab. We classified cells with 𝜈𝜈p =< 0.051 as significantly modulated 975 (Figure S3c).

125 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 36

976 Vibrissa motor cortex microstimulation

977 Animals were surgically prepared and habituated to head-fixation as described above. The 978 microstimulation72 was done with 0.3 ms, 50µA unipolar negative-tip current pulses at 100 979 Hz through a tungsten microelectrode in deep layers (putative layer 5) of the VMC (depth = 980 1500 µm from the dura). Current pulses were delivered from a stimulus isolator (World preci- 981 sion instruments, Sarasota, USA), gated by TTL pulses sent from a CED Power1401 by pro- 982 tocols written in Spike2 (Cambridge Electronic Design, Cambridge). The stimulation para- 983 digm was blocks of 1 s long stimulation trains interspersed with 1 s long pauses in stimulation 984 (”sham stimulation”). We observed the rats, and when the rats were whisking, we performed 985 the microstimulation protocol during the ongoing whisking. A random number generator en- 986 sured that the stimulation would either start with stimulation or with sham stimulation, as not 987 to bias the experiment. Synchronized 250 Hz digital high-speed video was recorded by trig- 988 gering frames and recording from a Spike2 script. The whisker movements were tracked us- 989 ing a custom written computer software for whisker tracking (Viktor Bahr, adapted from 990 Clack et al. 2012). Whisker tracking was done blind to the timing of stimulation and sham 991 stimulation.

992 To quantify the change in whisking power due to microstimulation, we filtered the trace to 993 remove jitter and calculated the whisking power as described above. To quantify the change 994 in whisker set angle during the 1 s stimulation periods, and 1 s sham periods, we fitted straight 995 lines to the whisker trace during each 1 s period. We took the slope of these straight lines to 996 be a measure of the average change of whisker set angle per time (°/s), and averaged across 997 these slopes for each experimental session.

998 For microstimulation in awake, socially interaction rats in the social gap paradigm, we used 999 the same microstimulation train as above, but the microstimulation was applied to deep layer 1000 VMC through the tetrode wires, implanted for recording (see above). Stimulation sites were 1001 conformed post-hoc by histology. In these experiments, the stimulation was triggered by the 1002 experimentor, who was watching the infrared video whenever the rats started socially interact- 1003 ing. The duration of social facial touch was quantified from the 25 frames/s video from the 1004 first to last whisker touch of the implanted rat. Since we had a varying number of data points 1005 pr. rat (depending on how many days the stimulation sites were found to be in VMC layer 5), 1006 we compared the differences between median social facial interaction duration between the 1007 stimulated and sham stimulated by fitting a linear mixed effects model (LME model) assum- 1008 ing a gaussian error distribution, with a random rat-specific intercept (‘Length ~isStim + (1| 1009 Rat)’) to account for mean differences among rats73.

1010 Vibrissa motor cortex blockade in awake rats

1011 Animals were surgically prepared and habituated to head-fixation as described above. Boro- 1012 silicate injection pipettes (Hirschmann Laborgeräte, Eberstadt, Germany) were pulled to an 1013 sharp tip and backfilled with Ringer or a 2% lidocaine solution74 (bela-pharm, Vechta, Ger- 1014 many). 250 nL Lidocaine was slowly pressure-injected into deep layers (putative layer 5) of 1015 the VMC (depth = 1500 µm from the dura) at two injection sites: (1.75 mm anterior, 1.5 mm 1016 lateral to bregma) and (1.25 mm anterior, 1.5 mm lateral to bregma), ~2-5 min pr. injection. 1017 Based on measurements on the spatial spread of injection of 2% lidocaine in cortex74, we es- 1018 timate that the injection of 250 nL lidocaine inactivated an area around the injection site de- 1019 fine by a sphere with a radius of 390 µm (which is given simply by the volume equation of 1020 the sphere . 250 Hz digital video was recorded by triggering ! ! (𝑉𝑉!"#!"#$%"&' ≈ !𝜋𝜋𝑅𝑅!"#$%!&#%'()

126 Ebbesen (2017) VMC activity suppresses contralateral touch

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 37

1021 frames and recording from a Spike2 script and the whisker movements were tracked using a 1022 custom written computer software for whisker tracking (Viktor Bahr, adapted from Clack et 1023 al. 201271). The whisking was filmed just following the lidocaine injections (i.e. in the few 1024 minutes range), where the inactivation by 2% lidocaine injection is largest, and before the cell 1025 activity recovers (which happens slowly in the 10-40 min post injection range)74. Whisker 1026 tracking was done blind to the injected solution (Ringer or lidocaine).

1027 The whisker trace was filtered, and whisking power and whisker set angle was calculated as 1028 described above. The ratio of the contralateral whisking power to the ipsilateral whisking 1029 power was found to be log-normally distributed (assessed with a Lilliefors test), so we per- 1030 formed log-normal t-tests instead of non-parametric tests to assess statistical significance of 1031 the ratios.

1032 Vibrissa motor cortex blockade in anaesthetized rats

1033 Rats were anaesthetized and prepared for head-fixation as described above. A square craniot- 1034 omy was microdrilled above VMC, 0.5 mm - 4.5 mm anterior bregma and 0.5 mm - 2.0 mm 1035 lateral. After dura removal, VMC was superfused with 30 µL blocking solution. The blocking 1036 solution was made from 500 µL 1 mM APV suspended in 0.1 M PBS, 50 µL 100 µM NBQX 1037 suspended in 0.1 M PBS and 500 µL Ringer’s solution. After superfusion of blocking solu- 1038 tion, the rat was intraperitoneally injected with 0.1 mL 0.1 mg/mL acepromazine. The animal 1039 was monitored until anaesthesia was light and whisker micromovements were observed (typi- 1040 cally ~ 60 mins post first ketamine/xylazine dose), and kept in light anaesthesia by additional 1041 alternating doses of 5% of the initial dose in ketamine/xylazine amount or 5% of the corre- 1042 sponding ketamine dose alone, respectively. As soon as whisker movements were observed, 1043 250 Hz high-speed videos of the whiskers were recorded at 10 min intervals until 160 mins 1044 post blocking. In one rat, the time course of the blocking of excitatory transmission in VMC 1045 was monitored with a field electrode, and found to be extinguished in deeper cortical layers of 1046 VMC ~ 100 mins post blocking. Whisker movements were tracked from the 250 Hz video 1047 and analyzed as described above.

1048 Vibrissa motor cortex blockade in anaesthetized mice

1049 Mice were anaesthetized and prepared for head-fixation as described above, but given 100 1050 mg/kg ketamin, 15 mg/kg xylazine. A square craniotomy was microdrilled above one hemi- 1051 sphere centered on VMC, 0.8 mm anterior bregma, 1.0 mm lateral. The mice supplemented 1052 with 0.01, 0.2 mg/mL acepromazine and a head-fixation post was applied to the skull with 1053 cyanoacrylate glue. The mice were head-fixed and kept at body temperature with a heating 1054 pad. We waited until the anaesthesia became light and we saw whisker movements begin to 1055 emerge, then VMC activity was blocked by injection of 25 mM muscimol (a GABAA receptor 1056 agonist, Sigma-Aldrich) suspended in Ringer’s solution at 10 nL/min (Huber et al. 2012) us- 1057 ing a QSI stereotactic injector (Stoelting). In two mice, we only blocked deep VMC, by an 1058 injection of 50 nL muscimol solution 900 µm below the dura, an another two mice, we 1059 blocked both superficial and deep VMC by injecting 100 nL muscimol solution at 900 µm 1060 and another 50 nL muscimol solution at 500 µm. Injection pipettes (Drummond 5µL) were 1061 labeled with DiI and the injection sites were confirmed to be VMC be perfusing the mice and 1062 locating the DiI-labeled pipette tracks by fluorescence microscopy. Whisker movements were 1063 tracked from the 250 Hz video and analyzed as described above.

1064

127 Ebbesen (2017) Chapter 7

Ebbesen et al. Vibrissa motor cortex suppresses contralateral touch 38

1065 Methods references

1066 65. Rao, R. P., Mielke, F., Bobrov, E. & Brecht, M. Vocalization–whisking coordination 1067 and multisensory integration of social signals in rat auditory cortex. Elife 3, 1–20 1068 (2014). 1069 66. Stockwell, R. G. Localization of the Complex Spectrum: The S Transform. IEEE 1070 Trans. Signal Process. 44, 998–1001 (1996). 1071 67. Schmitzer-Torbert, N., Jackson, J., Henze, D., Harris, K. & Redish, a. D. Quantitative 1072 measures of cluster quality for use in extracellular recordings. Neuroscience 131, 1–11 1073 (2005). 1074 68. Barthó, P. et al. Characterization of neocortical principal cells and interneurons by 1075 network interactions and extracellular features. J. Neurophysiol. 92, 600–608 (2004). 1076 69. Vigneswaran, G., Kraskov, A. & Lemon, R. Large Identified Pyramidal Cells in 1077 Macaque Motor and Premotor Cortex Exhibit ‘Thin Spikes’: Implications for Cell Type 1078 Classification. 31, 14235–14242 (2011). 1079 70. Houweling, A. R., Doron, G., Voigt, B. C., Herfst, L. J. & Brecht, M. Nanostimulation: 1080 manipulation of single neuron activity by juxtacellular current injection. J. 1081 Neurophysiol. 103, 1696–704 (2010). 1082 71. Clack, N. G. et al. Automated tracking of whiskers in videos of head fixed rodents. 1083 PLoS Comput. Biol. 8, e1002591 (2012). 1084 72. Tehovnik, E. J., Tolias, A. S., Sultan, F., Slocum, W. M. & Logothetis, N. K. Direct 1085 and indirect activation of cortical neurons by electrical microstimulation. J. 1086 Neurophysiol. 96, 512–521 (2006). 1087 73. Aarts, E., Verhage, M., Veenvliet, J. V, Dolan, C. V & van der Sluis, S. A solution to 1088 dependency: using multilevel analysis to accommodate nested data. Nat. Neurosci. 17, 1089 491–6 (2014). 1090 74. Tehovnik, E. J. & Sommer, M. A. Effective spread and timecourse of neural 1091 inactivation caused by lidocaine injection in monkey cerebral cortex. J. Neurosci. 1092 Methods 74, 17–26 (1997). 1093 1094 1095

128 Ebbesen (2017) VMC activity suppresses contralateral touch

129 Ebbesen (2017) Chapter 8 8. Motor cortex – To act or not to act?

This manuscript was submitted as: Ebbesen, C.L., & Brecht, M. (2017) Motor cortex – To act or not to act? (submitted to Nature Re- views Neuroscience on 21.0.3.2017), subsequently published as Ebbesen, C.L., & Brecht, M. (2017) Motor cortex – To act or not to act? (2017) Nature Reviews Neuroscience 18(1):694-705- This is the author’s version of this work, reprinted with permission from Nature Pub. Group

Figure 11: Movement suppression is under-represented in depictions of motor cortex (a) A textbook illustration of the human ‘motor’ (red) and somatosensory (green) cortex after the intraoperative stimulation experiments by Penfield & Rasmussen. The textbooks interpret human motor cortex as a ‘motor homunculus’, that is a muclelotopic motor map: “Motor cortex: movement” (red underscore, added). (b) Intraoperative photograph of an actual human motor (red, color added) and somatosensory cortex (green, color added) mapping experiment, as reported in the famous book by Penfield & Rasmussen (1952). Right, we see what the patient reports, when site 13 and 14 (yellow arrow, mid- 130 dle of motor cortex) are stimulated: a clear suppression of movement. Permissions: (a) Adapted from David Heeger (2006): Perception Lecture Notes: The Brain, Department of Psychology, New York University, Fair use (b) Adapted from Penfield & Rasmussen 1952 / Ebbesen& Brecht 2017, Nature Pub. Group Ebbesen (2017) Motor cortex – To act or not to act?

2 Motor cortex – To act or not to act? 3

5 by

(1,2) (1,3) 7 Christian Laut Ebbesen and Michael Brecht

(1) 8 Bernstein Center for Computational Neuroscience Berlin, Humboldt-Universität zu Berlin, Berlin, Germany 9 10 (2)Berlin School of Mind and Brain, Humboldt-Universität zu Berlin, Berlin, Germany

(3) 11 NeuroCure Cluster of Excellence, Humboldt-Universität zu Berlin, Berlin, Germany

13 Correspondence should be addressed to M.B. ([email protected])

18 Article Format: Opinion

21 WORD COUNT:

22 Abstract: 240

23 Main text: 5870 24 25

131 Ebbesen (2017) Chapter 8

Abstract 26 The motor cortex is a large frontal structure in the cerebral cortex of eutherian mammals. A

27 vast array of evidence implicates the motor cortex in the volitional control of motor output, 28 but how does the motor cortex exert this ‘control’? Ideas regarding the motor cortex have 29 historically been shaped by the discovery of cortical ‘motor maps’, i.e. ordered 30 representations of stimulation-evoked movements in anaesthetized animals. Volitional 31 control, however, entails the initiation of movements as well as the ability to suppress 32 undesired movements and behave in a non-reflexive fashion. In this review, we highlight 33 classic and recent findings which emphasize that motor cortex neurons not only initiate 34 movement, but also contribute strongly to movement suppression. Motor cortical stimulation 35 in awake subjects often leads to movement arrest and motor cortical inactivation often 36 disinhibits movements which are normally suppressed. Similarly, there is an unusual 37 predominance of suppression of motor cortical population activity during movement and an 38 increase of motor cortical activity in tasks which require the withholding of motor output. 39 Thus, stimulation, recording and inactivation studies suggest that the motor cortex – at least in 40 some instances – exerts a negative control of movement. This type of control is rather 41 different from the representation in sensory cortices, where sensory stimuli almost invariably 42 drive population firing rate increases and sensations. Action suppression is critical for the 43 strategic planning of behavior and numerous observations suggest that motor cortical activity 44 contributes heavily to this important and understudied cognitive ability.

132 Ebbesen (2017) Motor cortex – To act or not to act?

Introduction 49 The discovery of motor cortex as a ‘motor map’

50 In the late 1860s, two young physicians conducted an experiment which defined our thinking

51 about the brain and about motor control. Fritsch and Hitzig did not have laboratory working 52 space and therefore went home, tied down their experimental animals on Fritsch’s wife’s 53 dressing table, and performed one of the greatest neurophysiological experiments of all times: 54 They used electrodes to electrically stimulate the surface of the cerebral cortex of anesthetized 55 dogs (Figure 1a, left). They adjusted stimulation currents to evoke a tickling sensation when 56 applied to their own tongues. When Fritsch and Hitzig applied the same currents to specific 57 sites in the frontal cortex of their experimental animals, something very spooky happened: 58 Currents evoked movements of the experimental animals, whereby the type of evoked 1 59 movement varied with the cortical location of the stimulation site (Figure 1a, right). The 60 discovery of this ‘motor map’ paved the way for modern thinking about the cerebral cortex. 2 61 Ferrier soon reproduced Fritsch and Hitzig’s results in monkeys and after years of careful 62 experiments it gradually became clear that the cortical motor representation is highly 3 63 somatotopic, with a fine-grained 2D map of the external body . In later work, it became clear 4 64 that long and more intense stimulation trains can activate very complex motor patterns . 65 Although early experiments demonstrated that surgically lesioning the frontal ‘motor sites’ 1 66 did not abolish movements , these experiments indicated that a prime role of this cortical area 67 must be movement generation, hence the name motor cortex.

68 The discovery of a ‘movement suppression map’ in human motor cortex

69 A major advance in our understanding of the motor cortex came when cortical mapping 5 70 experiments were done in humans. The Canadian neurosurgeons Penfield and Rasmussen 71 mapped the cortex in awake patients during surgeries. This approach has an enormous 72 advantage over anaesthetized animal experimentation: the experimenter can ask the patient 73 how the cortical stimulation feels. These experiments yielded two insights: First, they 74 confirmed that the human motor cortex contains a somatotopic map of the body, a motor 75 ‘homunculus’ (Figure 1b, left). Secondly, the human patients often reported that motor cortex 76 stimulation lead to movement inhibition and muscle relaxation. Upon stimulation of motor 77 cortex sites, patients felt a sense of paralysis and numbness focal to specific body parts 78 (Figure 1b, right, during motor cortex stimulation: ‘I could not do it […] it felt like a 79 paralysis’ – ‘Oh, my right hand. I couldn’t move it.’). The Penfield and Rasmussen 80

133 Ebbesen (2017) Chapter 8

experiments demonstrated that the human motor cortex is not a pure ‘motor map’5, but also a 6 81 map of movement suppression .

82 Several subsequent studies have made similar observations and shown that so-called negative

83 motor areas (cortical areas where stimulation inhibits movement) are widespread across motor 7–10 11 84 cortex and premotor cortices . Such movement-suppressive effects of motor cortex 85 stimulation are impossible to discover in experiments on anaesthetized animals where only 86 ‘positive’ motor effects can be evaluated. Interestingly, we know from animal experiments 12–14 87 that many motor cortex stimulation sites produce no movements at all , but the ‘motor 88 maps’ observed from stimulating motor cortex in anesthetized animals have dominated our 15 89 thinking about motor cortex function . A further complication lies in the fact that ‘negative 90 motor area’ stimulation can be made to elicit positive movements (i.e. muscle twitches) with 10 91 an increasing stimulation current , and may thus remain unnoticed if only positive 11 92 stimulation effects are evaluated .

93 Is the motor cortex a brain structure for movement generation, a brain structure for

94 withholding movements – or both? Below, we review these two complementary views of 95 motor cortex function and highlight open questions. 96

134 Ebbesen (2017) Motor cortex – To act or not to act?

From motor cortex to muscle output 98 What do we mean, when we say ‘motor cortex’?

99 The motor cortex is a large frontal structure in eutherian mammals (Figure 2a). There are a

100 number of ways to define primary motor cortex. As described above, the ‘classic’ definition 101 of motor cortex is physiologically defined as the largest frontal area where e.g. 1,2,6,12,16 14,17 102 electrical or optogenetic stimulation elicits somatotopically organized movements 103 at low stimulation thresholds. In higher mammals, this ‘body-map’ of movements is adjacent 12,16,18–21 104 to and is a mirror image of the body map in primary somatosensory cortex . We 105 suggest that these four criteria (low stimulation thresholds for movements, full body 106 topography, adjacent to primary somatosensory cortex, mirror image topography of primary 107 somatosensory cortex) identify a homologous area in eutherian mammals. By this definition, 108 marsupials (an early branch of the mammalian tree) have only a large somatosensory 21–23 109 representation, but no motor cortex (Figure 2a, left). In rodents, the primary motor cortex 12,16,24,25 110 is very large and takes up almost all of frontal cortex (Figure 2a, middle). In primates, 111 frontal cortex contains several specialized premotor and prefrontal structures and the primary 112 motor cortex takes up (in relative terms), a much smaller area, a thin strip anterior to the 2,3,5 113 rolandic fissure (Figure 2a, right) .

114 Other definitions of motor cortex rely on anatomical markers such as a thick layer 5b and a 26,27 28,29 115 near-absent layer 4 , the frontal area of origin of corticospinal projections or the area of 30,31 116 dense corticocortical innervation from primary somatosensory cortex . Finally, some 32 117 definitions are based on mixed criteria and often rely on comparative anatomy to name 118 motor structures in e.g. the rodent brain after their putative corresponding primate 18,20,21,32 119 homologues . The precise correspondence between primate and rodent motor cortex 120 (and other premotor areas) is largely unknown and different ways of delineating motor cortex 33 121 sometimes suggest conflicting naming schemes . For example, the area of rat cortex which 12,16,17,19 122 the physiological approach designates as primary vibrissa motor cortex , is referred to 123 as secondary motor cortex (‘M2’, a putative homologue of primate supplementary motor 32 34–39 124 areas) by the rat brain atlas and some publications and as frontal orientation field 40–43 125 (‘FOF’, a putative homologue of primate frontal eye field) by others .

126 Parallel inhibitory and excitatory pathways from motor cortex to motoneurons

127 Motor cortical neurons do not innervate muscles directly. The most ‘direct’ pathways from

128 motor cortex to muscles are from so-called pyramidal tract-type neurons, which have their 129

135 Ebbesen (2017) Chapter 8

somata in layer 5 of motor cortex and send their axons through the pyramidal tract to target

130 neurons in the spine and brainstem. Some pyramidal tract neurons synapse directly onto 131 motor neurons which innervate muscles. This provides a straightforward circuit for motor 132 control: if pyramidal tract neurons spike, downstream motoneurons depolarize and spike, 44 133 which elicits muscle contraction . This direct wiring pattern in mammals, however, appears 45–49 44,50 134 to be the exception rather than the norm. In rodents as well as primates , the 135 predominant wiring pattern of corticobulbar and corticospinal projections from motor cortex 136 is not directly to motoneurons, but rather to brainstem and spinal interneurons, many of which 44,51 137 have inhibitory connections onto motor neurons (Figure 2b, left). The monosynaptic 138 projections from motor cortex to motor neurons in primates is a specialization of primate 139 distal limb muscles, which seems to have evolved for dexterous and fractionated digit 50,52–54 140 movements (Figure 2b, right).

141 Anatomical loops via other motor centers

142 In addition to direct corticospinal projections from motor cortex, there are major projections

143 from motor cortex to other cortical and subcortical motor structures, such as the 144 somatosensory cortex, the basal ganglia, the motor thalamus, the brain stem and the 55–57 145 cerebellum . Like the motor cortex, the somatosensory cortex also directly innervates 6,12,17,47 146 spinal and brain stem motor centers and can directly modulate muscle output . Many 147 motor cortical and pyramidal tract neurons send axon collaterals through the striatum and 148 target neurons in the subcortical nuclei of the basal ganglia, which also contains circuits for 149 both facilitation and suppression of muscle output. The basal ganglia circuitry is complex and 58–61 150 not sharply dichotomous , but in the ‘classic’ model, the basal ganglia is separated into the 58,62 151 so-called ‘direct’ and ‘indirect’ pathways . The net effect of exciting striatal neurons 152 through the direct pathway is a disinhibition of spinal motor centers, while the net effect of 153 exciting neurons through the indirect pathway is an increase in inhibitory drive from the basal 154 ganglia to downstream spinal motor centers.

155 Facilitation and suppression of muscle activity by motor cortical spikes

156 Through direct projections from motor cortex to spinal neurons and through indirect

157 projections via other motor centers, parallel anatomical loops exist, by which motor cortical 158 activity might potentially facilitate or suppress muscle activity. Spike-triggered averaging of 159 muscle electromyography (EMG) signals reveals that spikes of single motor cortical 44,63–65 160 pyramidal-tract neurons can predict both EMG peaks and EMG troughs . In primates, 161 monosynaptic excitatory connections between motor cortex neurons and spinal motoneurons 162

136 Ebbesen (2017) Motor cortex – To act or not to act?

are abundant, and net inhibitory connections to motoneurons are multisynaptic via spinal

163 interneuron microcircuits. Accordingly, peaks in the spike-triggered EMG (indicating muscle 164 facilitation) are abundant (~24% of neurons), large and have a short latency. Troughs in the 165 spike-triggered EMG, on the other hand, appear fewer in number (~2% of neurons) and are 166 smaller, presumably due to the temporal noise induced by the many synapses between motor 167 cortical spike and muscle contraction. This higher noise makes it likely that spike-triggered 168 averaging underestimates the functional connectivity of motor cortical neurons that lead to net 63–65 169 suppressive effects . More generally, spike-triggered averaging techniques are only well 170 suited to reveal oligosynaptic connections from motor cortex to motoneurons. During active 30 66 171 behavior, it is possible that motor cortex may act to initiate or to suppress motor programs 49,67,68 172 initiated reflexively from subcortical circuits including the basal ganglia and brainstem .

173 An alternative way of revealing the impact of motor cortical activity on muscle output is to

174 relate EMG signals to intra-cortical microstimulation. Motor cortex microstimulation also 175 causes increases and decreases in EMG signals and in the membrane potential of spinal 64,65,69–71 176 motorneurons (Figure 2c). In microstimulation experiments using brief stimulation 177 trains (~few ms), net EMG-suppression is much more common than that EMG- 64,65 178 facilitation , thus implicating motor cortical neurons in the suppression of muscular 179 activity. In microstimulation experiments using longer stimulation trains (several hundred 180 ms), it is possible to elicit coordinated sequences of muscle activation and inhibition towards 4,72 181 a range of body postures .

182 Clearly, analyzing the relationship between motor cortex activity and muscle activity by either

183 spike-triggered averaging techniques or by cortical microstimulation leads to very different 184 conclusions. Intracortical microstimulation induces neural activity that is very different from 73–75 185 ‘natural’ physiological patterns . Thus, the fact that investigations of the functional 186 connectivity from motor cortex to muscles suggest a substantial movement-suppressive role 187 of motor cortex activity might be viewed simply as artificial effects, which are interfering 188 with natural motor programs and have little physiological relevance. There is also evidence to 71 189 suggest that the relationship between motor cortex and muscle activity is highly dynamic . 190 Additionally, even though the effects of microstimulation point to a substantial role of motor 191 cortical activity in the suppression of muscle activity, suppression of muscle activity is not 192 necessarily suppression of motor output. Initiation of most motor actions, such as reaching, 193 involves both muscle excitation and inhibition. In the limb motor system there are agonist and 194 antagonist muscles that span the various joints which must be coordinated to make a 65,76 195 movement . 196

137 Ebbesen (2017) Chapter 8

Motor cortex – to act? The physiology of action 197 In this section, we will review motor cortex activity patterns which are associated with action.

198 Here, it is important to emphasize the immense diversity of motor cortex discharge patterns. 199 Such response diversity might contribute to behavioral flexibility of motor outputs, but it 200 limits the validity of general statements about motor cortex activity. In motor cortex – much 201 like in other cortices – cellular responses greatly vary between cortical layers and cell types; 202 the need for cell-specific readouts and unbiased population analysis is becoming increasingly 203 recognized.

204 Distal limb movements in monkeys

205 A major share of what we know about how motor cortical activity correlates with movement

206 comes from single-cell recordings in primates performed with reaching movements and hand- 207 manipulations. Just as with the interpretation of stimulation effects, it is a caveat that an 208 increase in motor cortical activity can be seen as facilitating movement in reference to the 209 excitation of agonist muscles but also as inhibition in reference to the antagonists. However, 210 even though some motor cortical neurons decrease their activity during such movements, the 211 majority of motor cortical and pyramidal-tract neurons correlate positively with movement 77–82 212 and force (Figure 3a). Motor cortex recordings during arm movements in monkeys 213 displayed peak firing rates just prior to movement onset and a cessation of activity before 77 214 movement completion, suggesting a role of motor cortex in movement initiation .

215 Complex activity patterns during limb movements in non-primates

216 In other mammals, the relationship of motor cortical firing rates with limb movements is

217 complex. In cats, for example, there is no overall modulation of motor cortical activity during 83,84 218 locomotion compared to rest . However, motor cortical activity increases with the 219 ‘difficulty’ of the locomotion, when the animal must make precise, controlled steps, such as 84–86 220 over obstacles or onto narrow steps of a ladder . Most motor cortical neurons 221 progressively increase their firing rate when the cat walks progressively slower to take smaller 87 222 steps between barriers . Motor cortical activity in cats is also related to phases of the step 223 cycle, with more spiking during the swing phase (where the foot is not touching the ground) 84–87 224 than during the stance phase (where the foot is applying force against the ground) .

225 With current techniques, it is now possible to investigate the relationship between motor 88 226 cortical activity and motor output with high fidelity in rodents . In contrast to the archetypal 227

138 Ebbesen (2017) Motor cortex – To act or not to act?

increase of motor cortical activity with movement observed in most studies on primate motor 77–82 228 cortex , there is a surprising predominance of suppression of rodent motor cortical activity 229 during movement across several studies and several motor behaviors. For example, a recent 230 study used extracellular recordings to investigate activity in deep layers of motor cortex in 231 mice running freely on a treadmill. During locomotion, the average spike rate of neurons in 232 deep layers of motor cortex decreased by 30% and single units, which discharged less spikes 233 during locomotion (66% of neurons), were much more common than neurons which increased 89 234 their firing rate during locomotion (34% of neurons) . Further, the reduction in spike rate 90 235 correlated with the spike width, such that units with wider spikes (putative principal cells , 91–93 89 236 see ) showed the strongest suppression of activity .

237 In a similar study, the activity of layer 5b neurons during locomotion in mouse motor cortex 94 238 was investigated with intracellular recordings . In this study, there was also an unusual 239 abundance of layer 5b motor cortical neurons which decreased their spike rate during 240 locomotion (Figure 3c). Overall, there was no modulation of motor cortical firing, but at the 241 single-cell level, there was a higher number of significantly activated (53%) compared to 94 242 suppressed (38%) layer 5b neurons during locomotion .

243 Vibrissa motor cortex activity decreases during whisking

244 In the rodent whisker system, movements can be easily quantified. Similar to locomotion, but

245 different from grasping and reaching, whisking movements are mediated by a subcortical 246 central pattern generator. The whisker motor plan is laid out for highly controlled whisker 95 96,97 247 protraction , which the animal uses to palpate objects in front of the nose . Vibrissa motor 248 cortex is huge (Figure 2a) and makes up approximately 6.5% of the whole cortical 12,16,19,24,25 249 sheet . Remarkably, vibrissa motor cortical neurons decrease spiking activity during 250 various whisking behaviors. Several studies have examined the relationship between vibrissa 67,98–100 251 motor cortex activity and whisking kinematics . The exact relation between vibrissa 252 motor cortex activity and whisker movement is still debated. An early study examined the 98 253 relationship between vibrissa motor cortex activity and whisker pad EMG and did not find 101 254 “any obvious correlation” . Several subsequent studies made similar observations and 67,98–100 255 found no overall modulation of vibrissa motor cortex population activity by whisking , 30,66 256 but this view has been challenged by two recent publications . One investigation found that 257 during whisking in head-fixed mice, there was a decrease in the firing rate of layer 2/3 and a 30 258 more mixed response pattern in layer 5 . Specifically, there was a small increase in layer 5 259 activity around the onset of whisking and single cells displayed both firing rate decreases and 260

139 Ebbesen (2017) Chapter 8

increases during whisking30. Similar to earlier conclusions from monkey motor cortex, such 30 261 findings might indicate a role of vibrissa motor cortex in whisking initiation . Another study 262 also investigated activity in layer 5 of vibrissa motor cortex, but this time in freely moving 263 rats during various forms of more naturalistic whisking behaviors. Exploratory whisking in 264 air, whisking to palpate objects and social whisking during facial interactions with 66 265 conspecifics were associated with ~21% overall decrease in spike rates in layer 5 of vibrissa 266 motor cortex. Further, intracellular recordings from layer 5 vibrissa motor cortex neurons in 267 socially interacting rats revealed that social whisking was associated with reduced cellular 66 268 excitability and membrane hyperpolarization .

269 Motor cortical inhibition during reaching movements

270 In a forelimb task, where head-fixed mice must push and pull a lever to receive a reward,

271 various studies have also revealed suppressed motor cortical activity during forelimb 272 movement: Across all layers of motor cortex, more putative pyramidal neurons were active 273 during phases of non-movement than during the actual movement phases. Based on 274 extracellular unit recordings across all layers, 43% of task-related neurons were active during 102 275 movement, while 57% were active during non-movement (Figure 3d). Juxtacellular 276 recordings targeted to deep-layer neurons (> 900 µm) yielded a more mixed picture with an 277 equal proportion of task-related neurons firing during movement and during non-movement. 278 Interestingly, however, the vast majority (94%) of fast-spiking interneurons in motor cortex 279 fired during the movement phase, suggesting that phases of forelimb movement are associated 102 280 with greatly increased motor cortical inhibition .

281 Another study investigated motor cortical activity in mouse motor cortex during reaches, 103 282 which were cued by a vibrotactile stimulus to the paw . The response of regular spiking 283 units during reaches was mixed with 20% of regular spiking units showing significantly 284 decreased and 39% showing significantly increased spike rates. In agreement with high levels 285 of motor cortical inhibition noted above, almost all parvalbumin positive interneurons 103 286 increased their firing rates .

287 In primates, attempts have also been made to investigate how motor cortical activity patterns

288 during reaching map onto principal and inhibitory cell types. One study used spike-width to 289 identify putative inhibitory interneurons in motor cortex and found that in this subpopulation, 290 firing rates tended to massively increase during movement, whereas the responses in putative 291 principal neurons was more mixed and showed less of an increase in firing rates upon 82 292 movement . It should be noted, however, that some primate pyramidal cells in macaque 293

140 Ebbesen (2017) Motor cortex – To act or not to act?

motor cortex have very narrow spikes104 and a large subset of interneurons in monkey motor 105 92 294 cortex have surprisingly wide spikes ; thus, this result should be taken with some caution .

295 Decreased corticospinal activity from rodent motor cortex during movement

296 We know relatively little about the activity patterns of identified corticospinal neurons in

297 rodent motor cortex during movement. Studies with recordings from identified neurons during 298 motor behavior have used immunohistochemistry techniques to assign projection targets to 66,94 299 recovered cells, but this approach yields relatively low cell numbers . Using retrograde 106 107 300 labeling techniques combined with extracellular recordings and calcium imaging , two 301 studies investigated the activity of pyramidal-tract projecting neurons in mouse anterior- 108,109 302 lateral motor cortex (a premotor structure involved in licking ) and primary motor cortex 303 during tongue movements. The activity of pyramidal tract projecting neurons displayed 304 complex temporal patterns in relation to tongue movements, showing both increases and 106,107 305 decreases with movement and during a delay period . These studies focused on 306 differences in firing rate between tongue movements in ipsilateral and contralateral directions. 307 While the population of pyramidal-tract neurons in anterior-lateral motor cortex as a whole, 106,107 308 across all firing patterns, showed a contralateral firing rate bias during movement , 309 pyramidal tract neurons in motor cortex showed no contralateral preference, but instead strong 107 310 somatosensory responses . Instead of comparing ipsilateral and contralateral trials, another 311 recent study used calcium-imaging of motor cortex layer 5b dendrites to investigate the 312 activity of identified corticospinal neurons in mouse motor cortex during a forelimb lever 313 press task. This study found that two-thirds of movement-related layer 5b corticospinal 110 314 neurons showed decreased activity during movement . 315

141 Ebbesen (2017) Chapter 8

Inactivation studies: Acting without motor cortex 316 Movement patterns after motor cortex inactivation

317 Even the first experiments by Fritsch and Hitzig demonstrated that surgically lesioning the 1 318 motor cortex did alter, but not abolish, movements in their experimental animals . In most 111 319 mammals the behavioral effects of motor cortical lesions are remarkably subtle . Many 112,113 320 simple behaviors (e.g. locomotion) persist even after total decortication due to the fact 321 that many complete neural circuits from sensory input to motor output are fully contained 114–116 322 within the spinal cord . For example, since locomotion is regulated by intrinsic pattern 114,115 323 generators in the spinal cord , a decorticate cat can display a wide range of natural gait 117,118 324 patterns when walking on a treadmill .

325 Symptoms of motor cortical lesions in humans and primates

326 Motor cortical lesions are associated with performance deficits in several movement related

327 tasks and massive ‘motor deficits’. The motor effects of motor cortical lesions greatly change 328 over time and, hence, are not easily categorized. For example, while the acute effects of motor 329 cortical lesions (muscle weakness and reduced, slowed-down movement) suggest that motor 330 cortex contributes positively to intact movement patterns, the chronic effects (spasticity, 119–121 331 clonus and hypertonia) point to a net loss of inhibitory control of motoneurons . A 332 prominent effect of motor cortical lesions in primates is a loss of fractionated movements and 333 the ability to independently move one body part without others: Attempts at individuated 334 movements of a given body part are accompanied by excessive, unintended motion of 119,122–124 335 contiguous body parts . These effects of motor cortex inactivation also suggest that a 336 prime role of descending motor cortex activity may be to ‘control’ – i.e. suppress and inhibit – 337 undesirable movements.

338 Non-primates: loss of movement suppression dominates

339 In many mammals, motor cortical inactivation leads to increases in movement. For example,

340 as perhaps suggested by the surprising abundance of motor cortical suppression during 341 whisking, locomotion and reaching (Figure 3b-d), the patterns of whisking and limb 342 movements after motor cortical inactivation reveal surprising deficits in movement 343 suppression and motor control.

344 Corresponding to the suppression of motor cortical activity during whisker movements 125 66 345 (Figure 3b), unilateral lesioning and unilateral inactivation of rat vibrissa motor cortex 346

142 Ebbesen (2017) Motor cortex – To act or not to act?

leads to protraction of the contralateral whiskers and increases the whisking power

347 contralaterally (Figure 4a). Corresponding to the suppression of motor cortical activity during 348 locomotion on a treadmill (in some cells) (Figure 3c), forelimb motor cortex lesions disinhibit 349 limb movements: Swimming rats normally hold their forelimbs still and swim with only their 350 hindlimbs, but after a unilateral forelimb motor cortex lesion, they start swimming with the 126 127,128 351 contralateral forelimb as well (Figure 4b). After motor cortex lesions, cats and 129,130 352 rats perform poorly in tasks requiring reaching and gripping food rewards. 353 Corresponding to the prevalence of suppressed neurons in a forelimb reaching task (Figure 354 3d), these deficits do not arise because the animals move too little, but because they lose 355 individual digit movements (all digits move together) and because they cannot control their 356 forelimb movements and ‘over-reach’ too far past their intended targets (Figure 4c).

357 Clearly, the interpretation of lesion-induced changes in behavior is complex. Several recent

358 studies have used optogenetic techniques to rapidly and reversibly inactivate motor cortex 359 during motor behavior and studies have made mixed observations. For example, in contrast to 131 360 observations made after lesioning or pharmacological inactivation of vibrissa motor 66 361 cortex , optogenetic activation of inhibitory interneurons in motor cortex did not affect 30 362 whisker set angle and reduced whisking amplitudes . Similarly, in contrast to the 127–130 363 overreaching induced by pharmacological motor cortical inactivation , optogenetic 364 activation of inhibitory interneurons in motor cortex blocked initiation and freeze execution of 365 reaches during a trained reaching task, but left untrained forelimb movements unaffected and 132 366 had no effect on initiation or execution of tongue movements . Perhaps the discrepancies 367 between the apparent effects of motor cortical lesions and transient optogenetic inhibition can 368 be at least partly explained by acute effects stemming from the rapid perturbance of the 133 134 369 homeostasis , but this remains an open question .

370 Motor cortex lesions lead to impaired performance in several movement-related tasks. In

371 many mammals, these deficits are not due to inability to generate movement, but to a lack of 372 controlled movements and compromised movement inhibition. 373

374

143 Ebbesen (2017) Chapter 8

Motor cortex – not to act? 375 The physiology of action suppression 376 Motor cortical neurons have usually been studied during motor performance. However, a few

377 recent studies have investigated the activity of motor cortical neurons in animals performing 378 tasks where withholding movement was required.

379 Several studies have investigated motor cortices using head-fixed experimental paradigms,

380 where mice are trained to respond to sensory stimulation by tongue movements in a go/no-go 381 paradigm: The animal must lick in response to the stimulus (‘hit’) and withhold licking to 135,136 382 other stimuli (‘correct rejection’) (Figure 5a, left). In trials where the mouse licked after 383 the stimulus (‘hit trials’), some layer 5 motor cortical neurons (41% ‘enhanced neurons’, with 384 high baseline firing rates) increased their firing rate during licking, while others decreased 385 their firing rate (20% ‘suppressed neurons’, with lower baseline firing rates) (Figure 5a, 386 middle). However, when the authors investigated the activity in trials where the mouse did not 387 lick after the sensory stimulation (‘miss trials’), they found that while the ‘enhanced neurons’ 135 388 were still increasing their firing rate, there was no modulation of the ‘suppressed neurons’ 389 (Figure 5a, right). The responses of ‘enhanced neurons‘ was highly correlated with the 390 sensory stimulation and only 500 ms after the stimulus (i.e. after or around the reaction time 391 of the mouse), there was also a small difference between ‘hit’ and ‘miss’ trials in the 135 392 ‘enhanced’ neurons . In other words, the ‘suppressed neurons’ in layer 5b of motor cortex 393 were strongly predictive of licking behavior (the actual motor output), whereas the ‘enhanced 394 neurons’ showed firing rate increases both when the mouse did and did not lick. It remains an 395 open question why ‘enhanced neurons’ relate so weakly to the actual movement output in 396 such a go/no-go task, while the suppressed neurons show a tight correlation with movement, 397 but it may be that enhanced neurons in motor cortex largely represent a sensory 107,135,136 137 398 signal , where a late component has been shown to correlate with perception .

399 Motor cortex inactivation and action suppression 400 In such sensorimotor go/no-go tasks, inactivation of sensory cortices leads to a degradation of

401 task performance because the licking ‘hit rate’ in go-trials is reduced. The animal correctly 402 does not lick to the no-go cue, but also stops licking to the go-cue (where he should lick), as if 403 he does not perceive the sensory stimuli. Motor cortex inactivation has the exact opposite 404 behavioral phenotype: When motor cortex is inactivated (Figure 5b, left), the ‘hit rate’ licking 405 remains high (the animal keeps licking to the go-cue) (Figure 5b, middle), but the animal also 406

144 Ebbesen (2017) Motor cortex – To act or not to act?

starts licking in no-go trials, where licking is supposed to be suppressed and is punished by 135,136 407 time-outs (Figure 5b, right). Behavioral performance in such go/no-to paradigms is often 408 quantified as d’ = Z(Hit rate)-Z(False alarm rate) and thus the inactivation of both motor 409 cortex and sensory cortex leads to a reduced ‘task performance’, but for opposite reasons.

410 The effect of optogenetic activation of inhibitory interneurons motor cortex mirror the effects

411 of pharmacological blockade: one study found that motor cortical inactivation did not stop 132 412 lick initiation or execution in a cued licking task and another study found that optogenetic 413 inactivation of motor cortex spared hit rate, but increased false alarm licking in a go/no-go 135 414 task .

415 The fact that motor cortex inactivation does not reduce ‘hit rate’ licking, but disinhibits

416 disadvantageous licking suggests that a prime role of motor cortex in such a task may not be 417 the generation of licking, but rather withholding tongue movements.

418 Withholding movement and waiting for rewards

419 Rats can learn to solve a task where they must initiate trials by poking in a nose-port, and then 35,36 420 wait for a reward, which arrives after a random time (Figure 5c, left). When the time to 421 reward is long, some rats will break the trial early and fail to receive the reward because of 422 their ‘impatience’. Recordings from motor cortex in rats solving such a ‘waiting task’ 423 revealed that there were more neurons which suppressed, rather than enhanced, their firing 35 424 rate when the rats moved away from the nose-port . Further, a large fraction (18%) of motor 425 cortical neurons showed activity just before or during the delay period, which was 426 significantly related to the time, which the rat decided to spend waiting for rewards. While the 427 response pattern of single neurons was mixed, the majority of these neurons showed positive 428 correlations between firing rate and waiting time, such that higher motor cortical firing 35 429 predicted longer waiting (Figure 5c, right). It should be added, however, that many motor 430 cortex cells showing delay activity also burst prior to movement onset. In a follow-up 431 investigation, the authors used demixed principal component analysis to interrogate motor 432 cortical activity for signals, which might be hidden in the ‘mixed’ population response. 433 Across the population, the pattern was the same: just before and during the waiting period, the 434 principal components of the population activity were positively correlated with waiting 36 435 time . Motor cortex inactivation disrupted performance in such a waiting task, for two major 436 reasons: First, most rats became ‘impatient’ during nose-poke trials and were not able to wait 437 long enough to receive the reward. Secondly, the rats spent more time moving, and less time 36 438 receiving the reward . 439

145 Ebbesen (2017) Chapter 8

It is a well-established finding across humans, monkeys and rodents that (pre)frontal cortical 111,138–140 440 lesions are associated with deficits in behavioral inhibition , which manifests itself in 441 impulsivity, “unrestrained and tactless” behavior and “fatuos jocularity and ill-timed bawdy 141 442 and puerile jokes” . In one rodent study, the authors examined how motor cortical activity is 142 443 modulated when behavioral inhibition is impaired due to inactivation of frontal cortex . Rats 444 were trained to perform a task, where they had to press a lever to initiate a trail, hold the lever 445 during a delay period of 1 second, and then release the lever to receive a reward. 446 Pharmacological inactivation of dorsomedial prefrontal cortex impaired the task performance, 447 due to a large increase in ‘premature responding’, i.e. the rats did not hold the lever down for 448 the full delay period, but released the lever too early and failed to receive the reward. 449 Interestingly, recordings from motor cortex revealed that the premature responding was not 450 associated with an increase in motor cortical activity during the delay period. Rather, the 142 451 inability to wait during the delay was associated with a decrease in motor cortical activity , 452 suggesting that motor cortical activity might be important for the suppression of premature 453 lever presses.

454 Motor cortex manipulation and waiting tasks

455 The interpretation that motor cortex plays an important role in the suppression of

456 disadvantageous lever presses is supported by a recent study. The study investigated the effect 457 of motor cortical inactivation in rats that were trained to press a lever twice to receive a 143 458 reward (Figure 5d, left). Both intact rats and rats with motor cortex lesions could learn to 459 solve the task and press the lever twice. Intact rats could learn to postpone the second lever 460 press to obtain a larger reward and once this had been learned, motor cortex ablation did not 461 affect the stereotyped/learned motor sequence, which the rats used to time their lever pressing. 462 Rats with motor cortex lesions, however, could not learn to postpone the second lever press 143 463 and continued to receive only low rewards by pressing the lever in fast succession (Figure 464 5d, right). 465

146 Ebbesen (2017) Motor cortex – To act or not to act?

Conclusions 466 Summary

467 We have reviewed findings suggesting that, in addition to movement generation, motor cortex

468 might contribute to movement suppression. The famous motor homunculus by Penfield & 469 Rasmussen is frequently presented as a movement map, but the observed stimulation effects 470 indicate a somatotopic organization of movement suppression (Figure 1). Motor cortical 471 pyramidal tract neurons are often presented as heavily innervating spinal motoneurons, but 472 this clear ‘movement circuit’ is an exceptional wiring pattern in mammals (Figure 2). In some 473 preparations, motor cortical activity increases upon movement, but movement is also 474 accompanied by a surprising prevalence of principal neuron suppression and increased motor 475 cortical inhibition (Figure 3). Motor cortical lesions interfere with, but do not abolish 476 movement and are often associated with impaired movement suppression (Figure 4). Motor 477 cortex has mainly been investigated as a structure for movement generation, but several 478 studies implicate motor cortex in the withholding of disadvantageous motor output (Figure 5).

479 A purely movement/action centered perspective does not capture motor cortical function

480 The activity patterns of motor cortical neurons during ongoing behavior are highly diverse. In 30,66,67,76,89,94,98,100,102 481 addition to relationships with movement , neurons in the primary motor 482 cortex have been implicated in choice, working memory and preparation of upcoming motor 40,43,144,145 483 decisions , in decision making in relation to rewards and upcoming motor 34–39 146–149 484 strategies , in ‘mirror neuron’-like representations of actions and in the 150 31,107,136,151 485 representation of visual and somatosensory stimuli. Such diversity is probably 107,136,152 486 functionally important and it emphasizes the importance of unbiased analysis. Thus, 487 rather than ‘searching’ for single cells with a-priori expected response pattern (e.g. positive 488 correlation with limb movement), we must also focus on analysis of e.g. multi-electrode or 489 imaging data to determine if systematic population responses exist and how archetypical 490 activity patterns map onto specific cell types and projection patterns.

491 Motor suppressive functions of motor cortices have received much less attention than the role 11 492 of motor cortex in movement generation, but these negative motor phenomena deserve 493 broader attention. Movement is often associated with a decrease in activity of principal 494 neurons, increases in activity of fast-spiking neurons and large amounts of movement-related 495 inhibition. This is an unexpected response pattern for a primary cortical area dedicated to 496 movement, since cortical neurons most commonly respond to relevant stimuli with increased 497

147 Ebbesen (2017) Chapter 8

activity. This is the case for sensory cortices153,154, and has been proposed as a general 155 498 principle for the flow of cortical information . There are many ways to reconcile decreases 76,81,156 499 in activity with a primarily movement-promoting function of motor cortical activity . 500 Nonetheless, because the population-firing-rate decrease to relevant stimuli is an unusual 501 cortical response pattern, this deserves more attention.

502 Behavioral inhibition across frontal cortex

503 Numerous animal studies111,157 and classic neuropsychological work (Phineas Gage158,159)

504 point to a major role of frontal and prefrontal cortex in the inhibitory control of 111,138–140 505 behavior . Frontal cortices are relatively large in primates while the primary motor 2,3,5 506 cortex is comparatively small , whereas in rodents, frontal cortex is almost entirely primary 12,16,19,24,25 507 motor cortex . In primates, several premotor structures have been shown to perform 508 movement-suppressive functions in the executive control of behavior. For example, the 509 primate frontal and cingulate cortex responses arising in the context of countermanding 160,161 510 occulomotor movement and antisaccades have been described . Similarly, both primate 511 studies and observations on human patients point to a major role of the supplementary motor 162–164 512 area in movement inhibition , and lesions to this area reveak involuntary, ‘alien’ 165 513 movements .

514 We wonder, if rodent motor cortex might be more general and more ‘frontal-like’ than the

515 potentially more movement-specialized primate motor cortex. Thus, while both the activity 516 patterns during movement and the movement patterns after cortical blockade lets it appear 517 likely that rodent motor cortex plays a major role in movement suppression, it needs to be 518 checked by comparative analysis if this is an archetypical feature of motor cortex. For 519 example, it would be interesting to see how marsupials perform on tasks of behavioral 166 520 inhibition, such as the classic marshmallow test .

521 Outlook: A strategic function of motor cortex

522 Volitional control of motor output means deciding when to move and when not to move. 167 523 Freud’s notion of Überich was based in a correct intuition about behavior: sometimes it is 524 very important to repress the urge to act on immediate desires. Action suppression is critical 525 to the strategic planning of motor behaviors, but we still know little about how motor cortex 526 contributes to this important cognitive capacity. We need to get away from a ‘movement’ 527 perspective, and further investigate motor cortex from a ‘behavioral strategy’ perspective. We 528 propose that future investigations of motor cortex function should study both movement and 529

148 Ebbesen (2017) Motor cortex – To act or not to act?

movement-suppression to elucidate how motor cortex allows mammals to behave in non-

530 reflexive ways. 531

149 Ebbesen (2017) Chapter 8

References 532 1. Fritsch, G. & Hitzig, E. Ueber die elektrische Erregbarkeit des Grosshirns. Arch. Anat. Physiol. wiss. Med. 37, 300–332 (1870). 533 534 2. Ferrier, D. Experiments on the Brain of Monkeys - No. I. Proc. R. Soc. B Biol. Sci. 23, 409–430 (1874). 535 536 3. Leyton, A. S. F. S. F. & Sherrington, C. S. S. Observations on the excitable cortex of the chimpanzee, orang-utan, and gorilla. Q. J. Exp. Physiol. 11, 135–222 (1917). 537 538 4. Graziano, M. S. a, Taylor, C. S. R. & Moore, T. Complex movements evoked by microstimulation of precentral cortex. Neuron 34, 841–851 (2002). 539 540 5. Asanuma, H. & Rosén, I. Topographical organization of cortical efferent zones projecting to distal forelimb muscles in the monkey. Exp. brain Res. 14, 243–56 541 (1972). 542 543 6. Penfield, W. & Rasmussen, T. The Cerebral Cortex of Man. (The Macmillan Company, 1952). doi:10.1212/WNL.1.1.96 544 545 7. Ikeda, A. et al. Movement-related potentials associated with bilateral simultaneous and unilateral movements recorded from human supplementary motor area. 546 Electroencephalogr. Clin. Neurophysiol. 95, 323–34 (1995). 547 548 8. Lüders, H. O., Dinner, D. S., Morris, H. H., Wyllie, E. & Comair, Y. G. Cortical electrical stimulation in humans. The negative motor areas. Adv. Neurol. 67, 115–29 549 (1995). 550 551 9. Nii, Y., Uematsu, S., Lesser, R. P. & Gordon, B. Does the central sulcus divide motor and sensory functions? Cortical mapping of human hand areas as revealed by electrical 552 stimulation through subdural grid electrodes. Neurology 46, 360–7 (1996). 553 554 10. Mikuni, N. et al. Evidence for a wide distribution of negative motor areas in the perirolandic cortex. Clin. Neurophysiol. 117, 33–40 (2006). 555 556 11. Filevich, E., Kühn, S. & Haggard, P. Negative motor phenomena in cortical stimulation: Implications for inhibitory control of human action. Cortex 48, 1251–1261 557 (2012). 558 559 12. Neafsey, E. J. et al. The organization of the rat motor cortex: A microstimulation mapping study. Brain Res. Rev. 11, 77–96 (1986). 560 561 13. Ayling, O. G. S., Harrison, T. C., Boyd, J. D., Goroshkov, A. & Murphy, T. H. Automated light-based mapping of motor cortex by photoactivation of 562 channelrhodopsin-2 transgenic mice. Nat. Methods 6, 219–24 (2009). 563 564 14. Harrison, T. C., Ayling, O. G. S. & Murphy, T. H. Distinct Cortical Circuit Mechanisms for Complex Forelimb Movement and Motor Map Topography. Neuron 565 74, 397–409 (2012). 566 567 15. Lemon, R. N. An enduring map of the motor cortex. Exp. Physiol. 93, 798–802 (2008).

568 16. Brecht, M. et al. Organization of rat vibrissa motor cortex and adjacent areas according to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells. 569 J. Comp. Neurol. 479, 360–373 (2004). 570 571

150 Ebbesen (2017) Motor cortex – To act or not to act?

17. Matyas, F. et al. Motor control by sensory cortex. Science 330, 1240–1243 (2010).

572 18. Molnár, Z. et al. Evolution and development of the mammalian cerebral cortex. Brain. Behav. Evol. 83, 126–39 (2014). 573 574 19. Hall, R. D. & Lindholm, E. P. Organization of motor and somatosensory neocortex in the albino rat. Brain Res. 66, 23–38 (1974). 575 576 20. Nudo, R. J. & Frost, S. B. 3 . 28 The Evolution of Motor Cortex and Motor Systems. (2007). 577 578 21. Kaas, J. H. Evolution of somatosensory and motor cortex in primates. Anat. Rec. - Part A Discov. Mol. Cell. Evol. Biol. 281, 1148–1156 (2004). 579 580 22. Frost, S. B., Milliken, G. W., Plautz, E. J., Masterton, R. B. & Nudo, R. J. Somatosensory and motor representations in cerebral cortex of a primitive mammal 581 (Monodelphis domestica): A window into the early evolution of sensorimotor cortex. J. 582 Comp. Neurol. 421, 29–51 (2000). 583 584 23. Karlen, S. J. & Krubitzer, L. The functional and anatomical organization of marsupial neocortex: Evidence for parallel evolution across mammals. Prog. Neurobiol. 82, 122– 585 141 (2007). 586 587 24. Gioanni, Y. & Lamarche, M. A reappraisal of rat motor cortex organization by intracortical microstimulation. Brain Res. 344, 49–61 (1985). 588 589 25. Zilles, K. & Wree, A. in The Rat Nevous System (ed. Paxinos, G.) 649–685 (Academic Press, 1995). 590 591 26. Brodmann, K. Vergleichende Lokalisationslehre der Großhirnrinde : in ihren Prinzipien dargestellt auf Grund des Zellenbaues. (J.A.Barth, 1909). 592 593 27. Yamawaki, N. et al. A genuine layer 4 in motor cortex with prototypical synaptic circuit connectivity. Elife 4, e05422 (2014). 594 595 28. Nudo, R. J. & Masterton, R. B. Descending pathways to the spinal cord, III: Sites of origin of the corticospinal tract. J. Comp. Neurol. 296, 559–583 (1990). 596 597 29. Rathelot, J.-A. & Strick, P. L. Muscle representation in the macaque motor cortex: An anatomical perspective. Proc. Natl. Acad. Sci. 103, 8257–8262 (2006). 598 599 30. Sreenivasan, V. et al. Movement Initiation Signals in Mouse Whisker Motor Cortex. Neuron 92, 1368–1382 (2016). 600 601 31. Ferezou, I. et al. Spatiotemporal Dynamics of Cortical Sensorimotor Integration in Behaving Mice. Neuron 56, 907–923 (2007). 602 603 32. Paxinos, G. & Watson, C. The rat brain in stereotaxic coordinates. (Academic Press, 1982). 604 605 33. Brecht, M. Movement, Confusion, and Orienting in Frontal Cortices. Neuron 72, 193– 196 (2011). 606 607 34. Barthas, F. & Kwan, A. C. Secondary Motor Cortex : Where ‘ Sensory ’ Meets ‘ Motor ’ in the Rodent Frontal Cortex. Trends Neurosci. 40, 181–193 (2017). 608 609 35. Murakami, M., Vicente, M. I., Costa, G. M. & Mainen, Z. F. Neural antecedents of self-initiated actions in secondary motor cortex. Nat. Neurosci. 17, 1574–1582 (2014). 610 611

151 Ebbesen (2017) Chapter 8

36. Murakami, M., Shteingart, H., Loewenstein, Y. & Mainen, Z. F. Distinct Sources of Deterministic and Stochastic Components of Action Timing Decisions in Rodent 612 Frontal Cortex. Neuron 94, 908–919.e7 (2017). 613 614 37. Reep, R. L., Goodwin, G. S. & Corwin, J. V. Topographic organization in the corticocortical connections of medial agranular cortex in rats. J. Comp. Neurol. 294, 615 262–280 (1990). 616 617 38. Reep, R. L., Corwin, J. V, Hashimoto, A. & Watson, R. T. Efferent connections of the rostral portion of medial agranular cortex in rats. Brain Res. Bull. 19, 203–21 (1987). 618 619 39. Sul, J. H., Jo, S., Lee, D. & Jung, M. W. Role of rodent secondary motor cortex in value-based action selection. Nat. Neurosci. 14, 1202–8 (2011). 620 621 40. Brody, C. D. & Hanks, T. D. Neural underpinnings of the evidence accumulator. Curr. Opin. Neurobiol. 37, 149–157 (2016). 622 623 41. Hanks, T. D. et al. Distinct relationships of parietal and prefrontal cortices to evidence accumulation. Nature 520, 220–223 (2015). 624 625 42. Erlich, J. C., Bialek, M. & Brody, C. D. A cortical substrate for memory-guided orienting in the rat. Neuron 72, 330–343 (2011). 626 627 43. Erlich, J. C., Brunton, B. W., Duan, C. A., Hanks, T. D. & Brody, C. D. Distinct effects of prefrontal and parietal cortex inactivations on an accumulation of evidence task in 628 the rat. 1–28 (2015). doi:10.7554/eLife.05457 629 630 44. Lemon, R. N. Descending Pathways in Motor Control. Annu. Rev. Neurosci 31, 195– 218 (2008). 631 632 45. O’Donoghue, D. L., Kartje-Tillotson, G. & Castro, a J. Forelimb motor cortical projections in normal rats and after neonatal hemicerebellectomy: an anatomical study 633 based upon the axonal transport of WGA/HRP. J. Comp. Neurol. 256, 274–283 (1987). 634 635 46. Grinevich, V., Brecht, M. & Osten, P. Monosynaptic pathway from rat vibrissa motor cortex to facial motor neurons revealed by lentivirus-based axonal tracing. J. Neurosci. 636 25, 8250–8258 (2005). 637 638 47. Sreenivasan, V., Karmakar, K., Rijli, F. M. & Petersen, C. C. H. Parallel pathways from motor and somatosensory cortex for controlling whisker movements in mice. Eur. 639 J. Neurosci. 41, 354–367 (2015). 640 641 48. Rouiller, E. M., Moret, V. & Liang, F. Comparison of the connectional properties of the two forelimb areas of the rat sensorimotor cortex: support for the presence of a 642 premotor or supplementary motor cortical area. Somatosens. Mot. Res. 10, 269–289 643 (1993). 644 645 49. Deschênes, M. et al. Inhibition, Not Excitation, Drives Rhythmic Whisking. Neuron 90, 374–387 (2016). 646 647 50. Kuypers, H. G. J. M. A New Look at the Organization of the Motor System. Prog. Brain Res. 57, 381–403 (1982). 648 649 51. Fetz, E. E., Perlmutter, S. I. & Prut, Y. Functions of mammalian spinal interneurons during movement. Curr. Opin. Neurobiol. 10, 699–707 (2000). 650 651 52. Rathelot, J.-A. & Strick, P. L. Subdivisions of primary motor cortex based on corticomotoneuronal cells. Proc. Natl. Acad. Sci. U. S. A. 106, 918–23 (2009). 652 653

152 Ebbesen (2017) Motor cortex – To act or not to act?

53. Heffner, R. & Masterton, B. Variation in form of the pyramidal tract and its relationship to digital dexterity. Brain. Behav. Evol. 12, 161–200 (1975). 654 655 54. Nakajima, K., Maier, M. A., Kirkwood, P. A. & Lemon, R. N. Striking differences in transmission of corticospinal excitation to upper limb motoneurons in two primate 656 species. J. Neurophysiol. 84, 698–709 (2000). 657 658 55. Osten, P. & Margrie, T. W. Mapping brain circuitry with a light microscope. Nat. Methods 10, 515–523 (2013). 659 660 56. Hooks, B. M. et al. Organization of cortical and thalamic input to pyramidal neurons in mouse motor cortex. J. Neurosci. 33, 748–60 (2013). 661 662 57. Jeong, M. et al. Comparative three-dimensional connectome map of motor cortical projections in the mouse brain. Sci. Rep. 6, 20072 (2016). 663 664 58. Calabresi, P., Picconi, B., Tozzi, A., Ghiglieri, V. & Di Filippo, M. Direct and indirect pathways of basal ganglia: a critical reappraisal. Nat. Neurosci. 17, 1022–1030 (2014). 665 666 59. Kress, G. J. et al. Convergent cortical innervation of striatal projection neurons. Nat. Neurosci. 16, 665–667 (2013). 667 668 60. Cui, G. et al. Concurrent activation of striatal direct and indirect pathways during action initiation. Nature 494, 238–42 (2013). 669 670 61. Tecuapetla, F., Jin, X., Lima, S. Q. & Costa, R. M. Complementary Contributions of Striatal Projection Pathways to Action Initiation and Execution. Cell 166, 703–715 671 (2016). 672 673 62. Grillner, S. & Robertson, B. The basal ganglia downstream control of brainstem motor centres—an evolutionarily conserved strategy. Curr. Opin. Neurobiol. 33, 47–52 674 (2015). 675 676 63. Cheney, P. D. & Fetz, E. E. Comparable patterns of muscle facilitation evoked by individual corticomotoneuronal (CM) cells and by single intracortical microstimuli in 677 primates: evidence for functional groups of CM cells. J. Neurophysiol. 53, 786–804 678 (1985). 679 680 64. Lemon, R., Muir, R. & Mantel, G. The effects upon the activity of hand and forearm muscles of intracortical stimulation in the vicinity of corticomotor neurones in the 681 conscious monkey. Exp. Brain Res. 66, 621–637 (1987). 682 683 65. Cheney, P. D., Fetz, E. E. & Palmer, S. S. Patterns of facilitation and suppression of antagonist forelimb muscles from motor cortex sites in the awake monkey. J. 684 Neurophysiol. 53, 805–820 (1985). 685 686 66. Ebbesen, C. L., Doron, G., Lenschow, C. & Brecht, M. Vibrissa motor cortex activity suppresses contralateral whisking behavior. Nat. Neurosci. 20, 82–89 (2017). 687 688 67. Hill, D. N., Curtis, J. C., Moore, J. D. & Kleinfeld, D. Primary motor cortex reports efferent control of vibrissa motion on multiple timescales. Neuron 72, 344–356 (2011). 689 690 68. Moore, J. D. et al. Hierarchy of orofacial rhythms revealed through whisking and breathing. Nature 497, 205–210 (2013). 691 692 69. Buys, E. J., Lemon, R. N., Mantel, G. W. & Muir, R. B. Selective facilitation of different hand muscles by single corticospinal neurones in the conscious monkey. J. 693 Physiol. 381, 529–49 (1986). 694 695

153 Ebbesen (2017) Chapter 8

70. Lemon, R. N., Mantel, G. W. & Muir, R. B. Corticospinal facilitation of hand muscles during voluntary movement in the conscious monkey. J. Physiol. 381, 497–527 (1986). 696 697 71. Davidson, A. G., Chan, V., O’Dell, R. & Schieber, M. H. Rapid changes in throughput from single motor cortex neurons to muscle activity. Science 318, 1934–7 (2007). 698 699 72. Ramanathan, D., Conner, J. M. & Tuszynski, M. H. A form of motor cortical plasticity that correlates with recovery of function after brain injury. Proc. Natl. Acad. Sci. U. S. 700 A. 103, 11370–11375 (2006). 701 702 73. Stoney, S. D., Thompson, W. D. & Asanuma, H. Excitation of pyramidal tract cells by intracortical microstimulation: effective extent of stimulating current. J. Neurophysiol. 703 31, 659–669 (1968). 704 705 74. Tehovnik, E. J., Tolias, A. S., Sultan, F., Slocum, W. M. & Logothetis, N. K. Direct and indirect activation of cortical neurons by electrical microstimulation. J. 706 Neurophysiol. 96, 512–521 (2006). 707 708 75. Histed, M. H., Bonin, V. & Reid, R. C. Direct Activation of Sparse, Distributed Populations of Cortical Neurons by Electrical Microstimulation. Neuron 63, 508–522 709 (2009). 710 711 76. Shenoy, K. V, Sahani, M. & Churchland, M. M. Cortical Control of Arm Movements: A Dynamical Systems Perspective. Annu. Rev. Neurosci. 36, 337–359 (2013). 712 713 77. Georgopoulos, a P., Kalaska, J. F., Caminiti, R. & Massey, J. T. On the relations between the direction of two-dimensional arm movements and cell discharge in primate 714 motor cortex. J. Neurosci. 2, 1527–1537 (1982). 715 716 78. Kalaska, J. F., Cohen, D. a, Hyde, M. L. & Prud’homme, M. A comparison of movement direction-related versus load direction-related activity in primate motor 717 cortex, using a two-dimensional reaching task. J. Neurosci. 9, 2080–2102 (1989). 718 719 79. Evarts, E. V. Relation of pyramidal tract activity to force exerted during voluntary movement. J. Neurophysiol. 31, 14–27 (1968). 720 721 80. Cheney, P. D. & Fetz, E. E. Functional classes of primate corticomotoneuronal cells and their relation to active force. J. Neurophysiol. 44, 773–791 (1980). 722 723 81. Fetz, E. E. Are movement parameters recognizable coded in the activity of single neurons? Behav. Brain Sci. 15, 679–690 (1992). 724 725 82. Kaufman, M. T., Churchland, M. M. & Shenoy, K. V. The roles of monkey M1 neuron classes in movement preparation and execution. J. Neurophysiol. 110, 817–825 (2013). 726 727 83. Armstrong, D. M. & Drew, T. Discharges of pyramidal tract and other motor cortical neurones during locomotion in the cat. J. Physiol. 346, 471–95 (1984). 728 729 84. Drew, T., Jiang, W. & Widajewicz, W. Contributions of the motor cortex to the control of the hindlimbs during locomotion in the cat. Brain Res. Rev. 40, 178–191 (2002). 730 731 85. Beloozerova, I. N., Farrell, B. J., Sirota, M. G. & Prilutsky, B. I. Differences in movement mechanics, electromyographic, and motor cortex activity between accurate 732 and nonaccurate stepping. J. Neurophysiol. 103, 2285–2300 (2010). 733 734 86. Drew, T. Motor Cortical Activity During Voluntary Cat . I . Cells Re lated to he Forelimbs Gait Modifications in the. J. Neurophysiol. 70, 179–99 (1993). 735 736

154 Ebbesen (2017) Motor cortex – To act or not to act?

87. Beloozerova, I. N. & Sirota, M. G. The role of the motor cortex in the control of vigour of locomotor movements in the cat. J. Physiol. 461, 27–46 (1993). 737 738 88. Ölveczky, B. P. Motoring ahead with rodents. Curr. Opin. Neurobiol. 21, 571–8 (2011). 739 740 89. Fisher, S. P. et al. Stereotypic wheel running decreases cortical activity in mice. Nat. Commun. 7, 13138 (2016). 741 742 90. Barthó, P. et al. Characterization of neocortical principal cells and interneurons by network interactions and extracellular features. J. Neurophysiol. 92, 600–608 (2004). 743 744 91. Gentet, L. J., Avermann, M., Matyas, F., Staiger, J. F. & Petersen, C. C. H. Membrane Potential Dynamics of GABAergic Neurons in the Barrel Cortex of Behaving Mice. 745 Neuron 65, 422–435 (2010). 746 747 92. Sjulson, L. L. Reforming our ideas about cell types and spike waveforms. J.Neurosci. (2011). at 749 750 93. Suter, B. A., Migliore, M. & Shepherd, G. M. G. Intrinsic electrophysiology of mouse corticospinal neurons: A class-specific triad of spike-related properties. Cereb. Cortex 751 23, 1965–1977 (2013). 752 753 94. Schiemann, J. et al. Cellular Mechanisms Underlying Behavioral State-Dependent Bidirectional Modulation of Motor Cortex Output. Cell Rep. 11, 1319–1330 (2015). 754 755 95. Dörfl, J. The musculature of the mystacial vibrissae of the white mouse. J. Anat. 135, 147–154 (1982). 756 757 96. Welker, W. I. Analysis of Sniffing of the Albino Rat. Behaviour 22, 223–244 (1964).

758 97. Sachdev, R. N. S., Sato, T. & Ebner, F. F. Divergent movement of adjacent whiskers. J. Neurophysiol. 87, 1440–8 (2002). 759 760 98. Carvell, G. E., Miller, S. A. & Simons, D. J. The relationship of vibrissal motor cortex unit activity to whisking in the awake rat. Somatosens. Mot. Res. 13, 115–127 (1996). 761 762 99. Friedman, W. A., Zeigler, H. P. & Keller, A. Vibrissae motor cortex unit activity during whisking. J. Neurophysiol. 107, 551–563 (2012). 763 764 100. Gerdjikov, T. V, Haiss, F., Rodriguez-Sierra, O. E. & Schwarz, C. Rhythmic whisking area (RW) in rat primary motor cortex: an internal monitor of movement-related 765 signals? J. Neurosci. 33, 14193–204 (2013). 766 767 101. Kleinfeld, D., Berg, R. W. & O’Connor, S. M. Anatomical loops and their electrical dynamics in relation to whisking by rat. Somatosens. Mot. Res. 16, 69–88 (1999). 768 769 102. Isomura, Y., Harukuni, R., Takekawa, T., Aizawa, H. & Fukai, T. Microcircuitry coordination of cortical motor information in self-initiation of voluntary movements. 770 Nat. Neurosci. 12, 1586–1593 (2009). 771 772 103. Estebanez, L., Hoffmann, D., Voigt, B. C. & Poulet, J. F. A. Parvalbumin expressing GABA-ergic neurons in primary motor cortex signal reaching. Cell Rep. (2017). 773 774 104. Vigneswaran, G., Kraskov, A. & Lemon, R. Large Identified Pyramidal Cells in Macaque Motor and Premotor Cortex Exhibit ‘Thin Spikes’: Implications for Cell Type 775 Classification. 31, 14235–14242 (2011). 776 777

155 Ebbesen (2017) Chapter 8

105. Zaitsev, A. V., Povysheva, N. V., Gonzalez-Burgos, G. & Lewis, D. A. Electrophysiological classes of layers 2-3 pyramidal cells in monkey prefrontal cortex. 778 J. Neurophysiol. 595–609 (2012). doi:10.1152/jn.00859.2011 779 780 106. Li, N., Chen, T., Guo, Z. V, Gerfen, C. R. & Svoboda, K. A motor cortex circuit for motor planning and movement. Nature (2015). doi:10.1038/nature14178 781 782 107. Chen, T.-W., Li, N., Daie, K. & Svoboda, K. A Map of Anticipatory Activity in Mouse Motor Cortex. Neuron 94, 866–879.e4 (2017). 783 784 108. Travers, J. B., Dinardo, L. A. & Karimnamazi, H. Motor and premotor mechanisms of licking. Neurosci. Biobehav. Rev. 21, 631–647 (1997). 785 786 109. Guo, Z. et al. Flow of cortical activity underlying a tactile decision in mice. Neuron 81, 179–194 (2014). 787 788 110. Peters, A. J., Lee, J., Hedrick, N. G., O’Neil, K. & Komiyama, T. Reorganization of corticospinal output during motor learning. Nat. Neurosci. (2017). doi:10.1038/nn.4596 789 790 111. Kolb, B. Functions of the frontal cortex of the rat: a comparative review. Brain Res. 320, 65–98 (1984). 791 792 112. Franz, S. I. & Lashley, K. S. The retention of habits by the rat after destruction of the frontal portion of the cerebrum. Psychobiology 3–18 (1917). 793 794 113. Semba, K. & Komisaruk, B. R. Neural substrates of two different rhythmical vibrissal movements in the rat. Neuroscience 12, 761–774 (1984). 795 796 114. Goulding, M. Circuits controlling vertebrate locomotion: moving in a new direction. Nat. Rev. Neurosci. 10, 507–518 (2009). 797 798 115. Kiehn, O. Locomotor circuits in the mammalian spinal cord. Annu. Rev. Neurosci. 29, 279–306 (2006). 799 800 116. Bourane, S. et al. Identification of a Spinal Circuit for Light Touch and Fine Motor Control. Cell 160, 503–515 (2015). 801 802 117. Whelan, P. Control of Locomotion in the Decerebrate Cat. Prog. Neurobiol. 49, 481– 515 (1996). 803 804 118. Graham Brown, T. The Intrinsic Factors in the Act of Progression in the Mammal. (1911). 805 806 119. Schieber, M. H. & Poliakov, A. V. Partial inactivation of the primary motor cortex hand area: effects on individuated finger movements. J. Neurosci. 18, 9038–9054 807 (1998). 808 809 120. Laplane, D., Talairach, J., Meininger, V., Bancaud, J. & Bouchareine, A. Motor consequences of motor area ablations in man. J. Neurol. Sci. 31, 29–49 810 811 121. Barnes, M. P. & Johnson, G. R. Upper Motor Neurone Syndrome and Spasticity. Clinical Management and Neurophysiology (Cambridge University Press, 2008). 812 813 122. Sasaki, S. et al. Dexterous finger movements in primate without monosynaptic corticomotoneuronal excitation. J. Neurophysiol. 92, 3142–3147 (2004). 814 815 123. Rouiller, E. M. et al. Dexterity in adult monkeys following early lesion of the motor cortical hand area: The role of cortex adjacent to the lesion. Eur. J. Neurosci. 10, 729– 816 740 (1998). 817 818

156 Ebbesen (2017) Motor cortex – To act or not to act?

124. Lawrence, D. G. & Kuypers, H. G. J. M. The Functional Organization of the Motor System in the Monkey. Brain 91, 1–14 (1968). 819 820 125. Gao, P., Hattox, A. M., Jones, L. M., Keller, A. & Zeigler, H. P. Whisker motor cortex ablation and whisker movement patterns. Somatosens. Mot. Res. 20, 191–198 (2003). 821 822 126. Stoltz, S., Humm, J. L. & Schallert, T. Cortical injury impairs contralateral forelimb immobility during swimming: a simple test for loss of inhibitory motor control. Behav. 823 Brain Res. 106, 127–32 (1999). 824 825 127. Martin, J. H. & Ghez, C. Impairments in reaching during reversible inactivation of the distal forelimb representation of the motor cortex in the cat. Neurosci. Lett. 133, 61–64 826 (1991). 827 828 128. Martin, J. H. & Ghez, C. Differential impairments in reaching and grasping produced by local inactivation within the forelimb representation of the motor cortex in the cat. 829 Exp. brain Res. 94, 429–43 (1993). 830 831 129. Castro, A. J. The effects of cortical ablations on digital usage in the rat. Brain Res. 37, 173–85 (1972). 832 833 130. Alaverdashvili, M. & Whishaw, I. Q. Motor cortex stroke impairs individual digit movement in skilled reaching by the rat. Eur. J. Neurosci. 28, 311–22 (2008). 834 835 131. Gao, P., Bermejo, R. & Zeigler, H. P. Whisker deafferentation and rodent whisking patterns: behavioral evidence for a central pattern generator. J. Neurosci. 21, 5374–80 836 (2001). 837 838 132. Guo, J.-Z. et al. Cortex commands the performance of skilled movement. Elife 4, 1–18 (2015). 839 840 133. Otchy, T. M. et al. Acute off-target effects of neural circuit manipulations. Nature 528, 358–363 (2015). 841 842 134. Südhof, T. C. Reproducibility: Experimental mismatch in neural circuits. Nature 528, 338–339 (2015). 843 844 135. Zagha, E., Ge, X. & McCormick, D. A. Competing Neural Ensembles in Motor Cortex Gate Goal-Directed Motor Output. Neuron 88, 565–577 (2015). 845 846 136. Huber, D. et al. Multiple dynamic representations in the motor cortex during sensorimotor learning. Nature 484, 473–478 (2012). 847 848 137. Sachidhanandam, S., Sreenivasan, V., Kyriakatos, A., Kremer, Y. & Petersen, C. C. H. Membrane potential correlates of sensory perception in mouse barrel cortex. Nat. 849 Neurosci. 16, 1671–7 (2013). 850 851 138. Dalley, J. W., Cardinal, R. N. & Robbins, T. W. Prefrontal executive and cognitive functions in rodents: Neural and neurochemical substrates. Neurosci. Biobehav. Rev. 852 28, 771–784 (2004). 853 854 139. Kim, S. & Lee, D. Prefrontal Cortex and Impulsive Decision Making. Biol. Psychiatry 69, 1140–1146 (2011). 855 856 140. Miller, E. K. The prefrontal cortex and cognitive control. Nat. Rev. Neurosci. 1, 59–65 (2000). 857 858 141. Mayer-Gross, W., Slater, E. & Roth, M. Clinical Psychiatry. (Cassell & Co., 1954).

859

157 Ebbesen (2017) Chapter 8

142. Narayanan, N. S. & Laubach, M. Top-Down Control of Motor Cortex Ensembles by Dorsomedial Prefrontal Cortex. Neuron 52, 921–931 (2006). 860 861 143. Kawai, R. et al. Motor Cortex Is Required for Learning but Not for Executing a Motor Skill. Neuron 86, 800–812 (2015). 862 863 144. Erlich, J. C., Bialek, M. & Brody, C. D. A Cortical Substrate for Memory-Guided Orienting in the Rat. Neuron 72, 330–343 (2011). 864 865 145. Hanks, T. D. et al. Distinct relationships of parietal and prefrontal cortices to evidence accumulation. Nature 520, 220–223 (2015). 866 867 146. Dushanova, J. & Donoghue, J. Neurons in primary motor cortex engaged during action observation. Eur. J. Neurosci. 31, 386–398 (2010). 868 869 147. Tkach, D., Reimer, J. & Hatsopoulos, N. G. Congruent activity during action and action observation in motor cortex. J. Neurosci. 27, 13241–13250 (2007). 870 871 148. Kraskov, a et al. Corticospinal mirror neurons Corticospinal mirror neurons. (2014).

872 149. Vigneswaran, G., Philipp, R., Lemon, R. N. & Kraskov, A. M1 corticospinal mirror neurons and their role in movement suppression during action observation. Curr. Biol. 873 23, 236–243 (2013). 874 875 150. Goard, M. J., Pho, G. N., Woodson, J. & Sur, M. Distinct roles of visual, parietal, and frontal motor cortices in memory-guided sensorimotor decisions. Elife 5, (2016). 876 877 151. Kleinfeld, D., Sachdev, R. N. S., Merchant, L. M., Jarvis, M. R. & Ebner, F. F. Adaptive filtering of vibrissa input in motor cortex of rat. Neuron 34, 1021–1034 878 (2002). 879 880 152. Kilner, J. M. & Lemon, R. N. What We Know Currently about Mirror Neurons. Curr. Biol. 23, R1057–R1062 (2013). 881 882 153. Mountcastle, V. Modalily and topographic properties of single neurons of cat’s somatic sensory system. J.Neurophysiol. 20, 408–434 (1956). 883 884 154. Hubel, D. H. & Wiesel, T. N. Receptive fields of single neurones in the cat’s striate cortex. J. Physiol. 148, 574–91 (1959). 885 886 155. Mountcastle, V. An organizing principle for cerebral function: the unit model and the distributed system. The Mindful Brain (MIT Press, 1978). 887 888 156. Sanes, J. N. & Donoghue, J. P. Plasticity and Primary Motor Cortex. Annu. Rev. Neurosci. 23, 393–415 (2000). 889 890 157. Laubach, M., Caetano, M. S. & Narayanan, N. S. Mistakes were made: Neural mechanisms for the adaptive control of action initiation by the medial prefrontal cortex. 891 J. Physiol. Paris 109, 104–117 (2015). 892 893 158. Harlow, J. M. Rocovery from the passage of an iron bar through the head. Publ. Massachusetts Med. Soc. 2, (1868). 894 895 159. Harlow, J. M. Passage of an iron rod through the head. Bost. Med. Surg. J. 39, 281–283 (1848). 896 897 160. Shadmehr, R. Distinct neural circuits for control of movement vs. holding still. J. Neurophysiol. jn.00840.2016 (2017). doi:10.1152/jn.00840.2016 898 899

158 Ebbesen (2017) Motor cortex – To act or not to act?

161. Schall, J. D., Stuphorn, V. & Brown, J. W. Monitoring and control of action by the frontal lobes. Neuron 36, 309–22 (2002). 900 901 162. Wardak, C. The Role of the Supplementary Motor Area in Inhibitory Control in Monkeys and Humans. J. Neurosci. 31, 5181–5183 (2011). 902 903 163. Nachev, P., Kennard, C. & Husain, M. Functional role of the supplementary and pre- supplementary motor areas. Nat. Rev. Neurosci. 9, 856–869 (2008). 904 905 164. Filevich, E., Kühn, S. & Haggard, P. Intentional inhibition in human action: The power of ‘no’. Neurosci. Biobehav. Rev. 36, 1107–1118 (2012). 906 907 165. Brainin, M., Seiser, A. & Matz, K. The mirror world of motor inhibition: the alien hand syndrome in chronic stroke. J. Neurol. Neurosurg. Psychiatry 79, 246–252 (2008). 908 909 166. Mischel, W., Ebbesen, E. B. & Raskoff Zeiss, A. Cognitive and attentional mechanisms in delay of gratification. J. Pers. Soc. Psychol. 21, 204–218 (1972). 910 911 167. Freud, S. Das Ich und das Es. Gesammelte Werke: XIII (Internationaler Psychoanalytischer Verlag, 1923). 912 913

914

915

159 Ebbesen (2017) Chapter 8

Acknowledgements 916 We thank James Poulet, Mikkel Vestergaard, Ann Clemens, Rajnish Rao and Andreea

917 Neukirchner for valuable discussions and comments on the manuscript. This work was 918 supported by Humboldt-Universität zu Berlin, BCCN Berlin (German Federal Ministry of 919 Education and Research BMBF, Förderkennzeichen 01GQ1001A), NeuroCure, and the 920 Gottfried Wilhelm Leibniz Prize of the DFG. 921 922 Competing interests statement

923 The authors declare no competing interests.

924

925

926

160 Ebbesen (2017) Motor cortex – To act or not to act?

Figures 927

928 Figure 1: Two opposing views of the motor cortex. (a) Discovery of motor cortex as a motor map. Left: Anaesthetized dog. Right: Sketch of 929 dog brain with indications of the cortical sites where stimulation evoked movements 930 (Fritsch and Hitzig)1. (Adapted with permission from ref. 1) 931 (b) Stimulation of human motor cortex often leads to movement suppression. Left: 932 Penfield and Rasmussen’s famous human motor homunculus. Right: Intra-operative 933 photograph and reports made by a patient during stimulation of motor cortical sites 934 (red color: M1, blue color: S1, white dots indicate rolandic and sylvian fissures)5. 935 (Adapted with permission from ref.5) 936 937

938

Permissions: (a) CC BY-NC 3.0 / Public domian (b) Macmillan / Public domain 161 Ebbesen (2017) Chapter 8

939 Figure 2: From motor cortex to muscle output: Anatomy and functional connectivity (a) Motor cortex is a large frontal structure in eutherian mammals. Left: Marsupials have 940 no motor cortex. Middle: In rodents, primary motor cortex takes up almost all of 941 frontal cortex. Right: In primates, frontal cortex is compartmentalized into specialized 942 pre-motor subfields (pale red), and the primary motor cortex (red) is comparatively 943 small. (Adapted with permission from refs. 21,33) 944 (b) Left: Old wiring scheme. In most animals, motor cortical axons terminate on spinal 945 interneurons, not directly on motoneurons (red dots indicate axons from M1). Right: 946 Recent wiring scheme (distal limbs in primates, larynx in humans): Some motor cortex 947 axons terminate directly on motoneurons.44,52. (Adapted with permission from ref. 44) 948 (c) Focal intra-cortical microstimulation reveals that motor cortical activity has both 949 facilitating, mixed and – most commonly –suppressive effects on muscular activity 950 (vertical lines indicate stimulation)64. (Reproduced with permission from ref. 64) 951 952

953

Permissions: (a,left) Wiley and Sons (a,mid,right) Elsevier (b) Annual Reviews (c) Springer 162 Ebbesen (2017) Motor cortex – To act or not to act?

954 Figure 3: Motor cortex – to act: Motor cortical activity during movement. (a) In primates, motor cortical firing rates increase with arm movements and force. Left: 955 In a classic experiment79, a monkey moves a lever connected to a weight in order to 956 receive a reward. Right: Motor cortical activity and arm trajectory when the monkey is 957 lifting a high weight (top) and no weight (bottom). (Adapted with permission from ref. 958 79) 959 (b) In rats, motor cortical activity decreases with whisking66. Left: A whisking rat. Right: 960 Peri-stimulus time histogram of a single unit in layer 5 of vibrissa motor cortex, 961 aligned to the beginning of whisking66. (Adapted with permission from ref. 66) 962 (c) In rats, motor cortical activity decreases with locomotion89,94. Left: A mouse running 963 on a treadmill. Right: Intra-cellular recording from neuron in layer 5b of motor cortex, 964 which is suppressed during locomotion94. (Adapted with permission from ref. 94) 965 966

Permissions: (a) Americal Phys. Soc. (b) own work / Nature Pub. Group (c) Elsevier (d) Nature Pub. Group 163 Ebbesen (2017) Chapter 8

(d) In rats, motor cortical activity decreases with reaching movements102. Left: A mouse performing a task, which requires pushing and pulling a lever. Middle: Identified 967 neuron in layer 5b of motor cortex. Right: Activity of the same neuron is suppressed 968 during reaching102. (Adapted with permission from ref. 102) 969 970

971

972

164 Ebbesen (2017) Motor cortex – To act or not to act?

Figure 4: Acting without motor cortex: Movement patterns after motor cortex 973 inactivation. 974 (a) Left: In intact rats, whisking is similar on both sides. Right: After unilateral vibrissa 975 motor cortex blockade, contralateral whisker move forward and contralateral whisking 976 power increases66,125. (Adapted with permission from ref. 66) 977 (b) Top: Swimming rats normally hold their forelimbs still and swim with only their 978 hindlimbs. Bottom: After a unilateral forelimb motor cortex lesion, rats start 979 swimming with the contralateral forelimb also126. (Adapted with permission from ref. 980 126) 981 (c) Left: Intact cats can be trained to reach for morsels of food inside small reaching 982 targets. Right: After forelimb motor cortex inactivation, cats fail to receive the rewards 983 because they move too much and over-reach the targets127. (Adapted with permission 984 from ref. 127) 985 986

Permissions: (a) own work / Nature Pub. Group (b) 165 Ebbesen (2017) Chapter 8

987 Figure 5: Motor cortex – not to act. The neurophysiology of not moving. (a) Left: Example of a ‘hit’ trial in a sensorimotor go/no-go task: after a whisker 988 stimulation, the mouse licks for a reward. Middle: In hit trials, a population of motor 989 cortical neurons increase their firing rate (‘enhanced neurons’, red line) and another 990 population decrease their firing rate (‘suppressed neurons’, blue line) upon stimulation 991 and during licking. Right: In ‘miss’ trials, where the whiskers are stimulated, but the 992 mouse does not lick, the population of ‘enhanced neurons’ respond nearly identically, 993 but there is no response in the population of ‘suppressed neurons’. (Adapted with 994 permission from ref. 135) 995 996

Permissions: (a,mid-right) Elsevier (b,mid-right) Elsevier (c) Nature Pub. Group (d) Elsevier 166 Ebbesen (2017) Motor cortex – To act or not to act?

(b) Left: Example of a ‘false alarm’ trial in a sensorimotor go/no-go task: the mouse licks in the absence of whisker stimulation and is punished by a time out period. Right: 997 Motor cortical inactivation does not affect hit rate (where the rat must lick), but 998 massively increases false alarm rate (where licking must be withheld)135,136 (Adapted 999 with permission from ref. 135) 1000 (c) Left: Rat waiting for a reward, which arrives after a random time. Right: Activity of a 1001 single motor cortical neuron while the rat is waiting. Longer waiting times (dark 1002 colors) are associated with higher motor cortical activity35,36. (Adapted with 1003 permission from ref. 35) 1004 (d) Left: Both intact rats and rats with motor cortex lesions can learn to solve a task, 1005 where they must press a lever twice to receive a reward. Right: Intact rats can learn to 1006 postpone the second lever press to receive a larger reward, but rats with motor cortex 1007 lesions cannot learn to postpone143. (Adapted with permission from ref. 143) 1008 1009

1010

167 Ebbesen (2017) Chapter 9 9. General Discussion 9.1 Parahippocampal cortex and neural circuits underlying spatial exploration 9.1.1 Structure-function relationships in parahippocampal cortex In the first part of this thesis I presented work, which showed that both spatial response patterns and temporal coding patterns map onto the anatomical structure of parahippocampal cortex with cell- type specificity. We found differences between neurons in parasubiculum and the superficial median entorhinal cortex (Chapters 5-6). Further, within the medial entorhinal cortex, we found differences between neurons in layer 3 and layer 2 (Chapter 5-6 and Tang et al., 2015). Finally, within layer 2 of the medial entorhinal cortex, we found differences between stellate neurons and pyramidal neurons (Chapter 4 and Reifenstein et al., 2016).

Before our work, it was already know that parahippocampal neurons show correlations between spatial and temporal response patters. For example, extracellular recordings had revealed that grid cells in layer 2 of medial entorhinal cortex were overwhelmingly theta-modulated neurons (Boccara et al., 2010). Based on intrinsic cell properties, stellate neurons show a higher tendency to discharge at theta frequencies in-vitro than pyramidal neurons (Alonso and Klink, 1993; Shay et al., 2015). Thus, based on in-vitro spike patterns and intrinsic properties, stellate neurons seemed the most likely candidates for grid cells (Latuske et al., 2015; Moser et al., 2008; Rowland et al., 2016). On the other hand, it was also known from extracellular recordings that grid cells were not uniformly distributed across layer 2 of the medial entorhinal cortex, but rather clustered in grid activity ‘hot-spots’ (Stensola et al., 2012) with a size and spacing corresponding to the anatomical clusters of pyramidal neurons (Ray et al., 2014). Thus, the anatomical distribution of grid cells rather pointed to the conclusion that grid cells might be pyramidal neurons (Brecht et al., 2014; Burgalossi and Brecht, 2014; Savelli and Knierim, 2014).

9.1.2 How are grid cells made? We could show that differences between temporal spiking patterns of stellate and pyramidal neurons (Ray et al., 2014) suggested that grid cells were primarily pyramidal neurons and that border cells were primarily stellate neurons (Chapter 4). Obviously, this conclusion is indirect and rests on assumptions about the stationarity of theta modulation within and between neurons (Chapter 4 & 6). Thus, even though our observations were quite clear, our method cannot exclude that e.g. a small subpopulation of stellate neurons spike with otherwise pyramidal-typic theta modulation and a spatial grid pattern.

168 Ebbesen (2017) General Discussion

Interestingly, despite massive efforts by several groups, we still have only little data indicating if such a stellate-grid-cell subpopulation exists, if grid cells are primarily pyramidal neurons, or if grid cells are equally prevalent in both cell types. The paucity of data mainly stems from the fact that grid cells are not very abundant (~ 5-15% of principal neurons, Chapter 4) and must be recorded in animals exploring a full two-dimensional environment to be unequivocally identified. These factors contribute to making grid cells highly challenging to study, and thus much of our knowledge about structure- function relationships in parahippocampal cortex must be taken with important caveats. For example, in addition to entorhinal grid cells being theta-modulated (Boccara et al., 2010), Chapter 4), two studies found that across parahippocampal cortex, grid cells were also most often bursty neurons (La- tuske et al., 2015; Newman and Hasselmo, 2014). According to our studies on cell-type differences in burstiness (Chapter 6), this suggests that parahippocampal grid cells are most likely either pyramidal neurons in layer 2 of medial entorhinal cortex or parasubicular neurons, with the caveat that we have to assume that temporal spike patterns map onto cell type in a well-behaved manner.

Another study made intracellular recordings from neurons in layer 2 of medial entorhinal cortex and assigned the neurons as pyramidal or stellate by their reconstructed morphology. This study found that stellate neurons, not pyramidal neurons, were most likely to be grid cells (Domnisoru et al., 2013). While the morphological and laminar assignment of these neurons was clear, the neurons were recorded in rats exploring a linear track (not a full two-dimensional environment). In such a setup, it is extremely difficult to distinguish a true grid cell from a neuron with another spatial, but ‘non-grid’ pattern (Yoon et al., 2016). Two recent studies used transgenic mice and viral techniques to restrict the expression of a calcium sensor and light-gated ion channels to pyramidal and stellate neurons. In one study, in-vivo imaging of calcium-activity suggested that the proportion of grid cells was remarkably similar in the two cell types (Sun et al., 2015). While this study had high certainty of the cellular identity (dentate gyrus-projecting ‘stellate’ versus wolframin-positive ‘pyramidal’ neurons), dorsal implantation of an endoscope for in-vivo imaging massively disrupts the tissue and impairs the ability to distinguish e.g. parasubicular and entorhinal neurons, which both express wolframin. Another study used optogenetic activation to identify extracellularly recorded stellate and pyramidal neurons and similarly found that grid cells were equally distributed across cell types (Toader, 2016), with the major caveat that due to the very low latencies of some synaptically activated neurons (Muñoz et al., 2014), such optogenetic tagging-studies are susceptible to high rates of false-positives, especially when optoge- netcally activating excitatory cell types (Toader, 2016). 169 Ebbesen (2017) Chapter 9

9.1.3 What does the grid cell system teach us about cortical computation? Numerous anatomical studies have clearly shown that the cytoarchitechtonic structure of parahippocampal cortex is highly modular (Burgalossi and Brecht, 2014), comprising both the hexagonal arrangement of patches of pyramidal neurons in layer 2 of the medial entorhinal cortex ((Kitamura et al., 2014; Naumann et al., 2016; Ray et al., 2014), perhaps with further subdivisions (Fuchs et al., 2016)), but also other anatomical modules within layer 2 (Ray et al., 2017), layer 3 (Henn-Mike et al., 2016; fnsys-11-00020 April 6, 2017 Time: 18:7 # 14 Ray et al., 2017; Tang et al., 2015) and the adjacent presubiculum (Preston-Ferrer et al., 2016; Ray et al., 2017) and parasubiculum (Chapter 5, (Burgalossi et al., 2011; Ray et al., 2017)) (Figure 12). Our Ray et al. Entorhinal Modules work suggests that both spatial and temporal temporal coding features are highly determined by these

anatomical modulesfnsys-11-00020 (Chapters April4,6). 6, 2017 Time: 18:7 # 14

fnsys-11-00020 AprilThere 6, 2017 is Time:a wealth 18:7 # 14of computationalRay et al. models, which propose mechanisms, which could generate spatial Entorhinal Modules discharge patterns in parahippocampal neurons (Giocomo et al., 2011; Zilli, 2012). Models where

Ray et al. the anatomical structure patterns the activity are rare (but see Brecht Entorhinalet al., Modules 2014) and most models propose mechanisms by which networks of similar (most often stellate) neurons organize to generate

PrS e.g. periodic grid-cell-like a Por PaS b spike patterns, commonly by oscillatory interference and/or by network at-

MEC tractor dynamics (but see e.g. Kerdels, 2016). Such

FIGURE 9 | Modular structures inmodels the parahippocampal do not region. Overviewin their of modular structures in the superficial layers of MEC and neighboring regions. Cb D (A) Overview of modular structures in the superficial layers of MEC and the neighboring, mMEC, PaS, and PrS. Sections stained for different markers were aligned at WFS1 L M the entorhinal/parasubicular border and modular structures drawn. (B) Schematic illustration of modular structures in the dorsal MEC and parasubiculum. D, dorsal; 500 µm V V, ventral; L, lateral; M, medial. © 2006present Elsevier, (B) adaptedform from accountWitter and Moser for (2006) . Adapted with permission from Elsevier. Figure 12: Modular cytoachitectonics of parahippocampal cortex. the stratified distribu- (a) Tangential section of rat parahippocampal cortex throughbe considered the aspre part- of the mMEC or PaS. Ontogenetically, recently, detailed studies using chemoarchitecture, molecular the PaS shows calbindin-expressiontion inof early parahippocampal development (Ray markers and single-axon tracing have revealed the modular subiculum (PrS), parasubiculum (PaS), postrhinal cortexand Brecht, (Por) 2016 and), and by maturation only a narrow stripe archictecture of the presubiculum in macaque monkeys in great layer 2 of the medial entorhinalFIGURE 9 | Modular cortex structures (MEC), in the parahippocampal stainedremains region.for at calbindin theOverview border. of modular However, structures thisin calbindin-stripe the superficial layers partially of MEC and lies neighboringdetail (regions.Ding et al., 2000; Ding and Rockland, 2001; Ding, 2013). (A) Overview of modular structures in the superficial layersoutside of MEC the and traditionalthe neighboring, boundaries mMEC,activity PaS, of and the PrS. PaS,onto Sections as determined stainedspecific for different by cel markersStudies- were on aligned the modularat structure of the rodent presubiculum (Cb, green channel) andthe wolframin entorhinal/parasubicular (WFS1, border and modularred channel). structuresAChE drawn. and (B)Clusters ZincSchematic histochemistry. illustration of of modular The mMEC structures also in the has dorsal a distinctive MEC and parasubiculum.have focused D, dorsal; on the development of cellular modules (Nishikawa V, ventral; L, lateral; M, medial. © 2006 Elsevier, (B) adapted from Witter and Moser (2006). Adapted with permission from Elsevier. calbindin-positive pyramidal neurons are visible in thelayout medial compared en to- the rest oflular the MEC, ensembles as determined with by the diset- al., 2002) and have shown cytochrome oxidase modules distribution of calretinin, calbindin, and parvalbumin positive in adult animals (Gonzalez-Lima and Cada, 1998) but are FIGURE 9 | Modular structures in the parahippocampal region. Overview of modular structures in the superficialcells layers ( ofFigure MEC and 3). neighboring Future functional, regions. ontogenetic, and evolutionary (A) Overview of modulartorhinal structures cortex. in the superficial (Adapted layers ofbe MEC considered andfrom the neighboring, Ebbesen as part mMEC, of theet PaS, mMECal., and PrS.2016, or Sections PaS. Chap stained Ontogenetically, for different4.) markersrecently, were aligned detailed at studies using chemoarchitecture,still relatively molecular scarce. Integrating data on parahippocampal studies would help us deciphertinct the role archetypal of this region. temporal the entorhinal/parasubicular(b) Schematic border and overview modular structures of drawn.the modular PaS(B) Schematic shows structures calbindin-expression illustration of modular in structuresthe in earlysuperficial in the development dorsal MEC andlayers (Ray parasubiculum. markers D, dorsal; and single-axon tracing have revealedstructure the from modular rodents, primate, and other mammals is critical V, ventral; L, lateral; M, medial. © 2006 Elsevier, (B) adapted from Witter and Moser (2006). Adapted with permission from Elsevier. for translational research (Ding, 2013; Naumann et al., 2016), and Brecht, 2016), and by maturation onlyOverview a narrow stripe of Modulararchictecturespike Structures of patterns the presubiculum in the(e.g. in bursty, macaque monkeys in great of MEC and the neighbouringremains at the medial-most border. However, this MEC calbindin-stripe (mMEC), partially PaS, lies however, here we narrowly focus on the rat parahippocampal Parahippocampaldetail Region (Ding et al., 2000; Ding and Rockland, 2001; Ding, 2013). outside the traditional boundaries of the PaS, as determined by Studies on the modular structure of the rodentregions. presubiculum Staining for acetylcholinesterase activity in retrosplenial be consideredand as part PrS. of the Colored mMEC or regions PaS. Ontogenetically, indicate anatomicalrecently, detailed ‘modules’ studies using of chemoarchitecture, neurons molecular AChE and Zinc histochemistry. The mMEC alsoCholinergic has a distinctive and zincergichave modules focusedtheta-rhythmic are on thenot developmentexclusive to pyramidal the of cellularMEC. modulesgranular (Nishikawa cortex and presubiculum reveals distinct clusters as the PaS shows calbindin-expression in early development (Ray markers and single-axon tracing have revealed the modular delineated by enrichmentlayout in compared synaptic to the Zinc rest of activity the MEC, as(Zinc),In determined several enriched adjacent by the areaset al., such 2002 as) retrosplenial and have shown granular cytochrome cortex in oxidase MEC but modules also reveal a number of differences: (i) Each area and Brecht, 2016), and by maturation only a narrow stripe archictecture of the presubiculum(Ichinohe, in 2012macaque) and monkeys presubiculum, in great acetylcholinesterase staining has a distinct density and distribution of acetylcholinesterase distribution of calretinin, calbindin, and parvalbumin positive in adultneurons animals ( Gonzalez-Limavs. non-bursty and Cada, 1998) but are remains at the border.cholinergic However, thisactivity calbindin-stripe (AChE) partially and lies expressiondetail (Ding etof al., calbindin 2000; Dingshows and (Cb), a Rockland,strikingly cal modular 2001- ; Ding, distribution, 2013). but at present it is not activity clusters. (ii) MEC does not show a matching M2 receptor cells (Figure 3). Future functional, ontogenetic, and evolutionary still relatively scarce. Integrating data on parahippocampal outside the traditional boundaries of the PaS, as determined by Studies on the modular structureknown how of the they rodent relate presubiculumto modular terminations of cholinergic distribution (data not shown; Rouse and Levey, 1996; Wang retinin (Cr) and parvalbuminstudies would (PV). help us (Adapted decipher the role from of this Ray region. et al., 2017,structure from rodents, primate, and other mammals is critical AChE and Zinc histochemistry. The mMEC also has a distinctive have focused on the developmentfibers inof cellular the entorhinal modules ( cortexNishikawastellate or elsewhere. neurons The in modular layer et2 al., 2011). (iii) Calbindin-positive cells in presubiculum form a for translational research (Ding, 2013; Naumann et al., 2016), layout comparedwith to the permission, rest of the MEC, Frontiers) as determined by the et al., 2002) and have shownstructure cytochrome of the presubiculum oxidase modules is most prominent in primates lattice surrounding acetylcholinesterase activity clusters, whereas Overview of Modular Structures in the however, here we narrowly focus on the rat parahippocampal distribution170 of calretinin, calbindin, and parvalbumin positive in adult animals (Gonzalez-Limaand humans and and Cada, was 1998 described) but in are classical studies of cellular there are few calbindin-positive cells in superficial layers of Parahippocampal Region regions. Staining for acetylcholinesterase activity in retrosplenial cells (Figure 3). Future functional, ontogenetic, and evolutionary still relatively scarce. Integratingarchitecture data (Rose, on parahippocampal1927; Altschul, 1933; Braak, 1978). More retrosplenial granular cortex. Cholinergic and zincergic modules are not exclusive to the MEC. granular cortex and presubiculum reveals distinct clusters as studies would help us decipher the role of this region. structure from rodents, primate, and other mammals is critical In several adjacent areas such as retrosplenial granular cortex in MEC but also reveal a number of differences: (i) Each area for translational research (Ding, 2013; Naumann et al., 2016), (Ichinohe, 2012) and presubiculum, acetylcholinesterase staining has a distinct density and distribution of acetylcholinesterase Overview of Modular Structures in the however, here we narrowlyFrontiers focus in onSystems the Neuroscience rat parahippocampal | www.frontiersin.org 14 April 2017 | Volume 11 | Article 20 shows a strikingly modular distribution, but at present it is not activity clusters. (ii) MEC does not show a matching M2 receptor Parahippocampal Region regions. Staining for acetylcholinesterase activity in retrosplenial known how they relate to modular terminations of cholinergic distribution (data not shown; Rouse and Levey, 1996; Wang Cholinergic and zincergic modules are not exclusive to the MEC. granular cortex and presubiculum reveals distinct clusters as fibers in the entorhinal cortex or elsewhere. The modular et al., 2011). (iii) Calbindin-positive cells in presubiculum form a In several adjacent areas such as retrosplenial granular cortex in MEC but also reveal a number of differences: (i) Each area structure of the presubiculum is most prominent in primates lattice surrounding acetylcholinesterase activity clusters, whereas (Ichinohe, 2012) and presubiculum, acetylcholinesterase staining has a distinct density and distribution of acetylcholinesterase and humans and was described in classical studies of cellular there are few calbindin-positive cells in superficial layers of shows a strikingly modular distribution, but at present it is not activity clusters. (ii) MEC does not show a matching M2 receptor architecture (Rose, 1927; Altschul, 1933; Braak, 1978). More retrosplenial granular cortex. known how they relate to modular terminations of cholinergic distribution (data not shown; Rouse and Levey, 1996; Wang fibers in the entorhinal cortex or elsewhere. The modular et al., 2011). (iii) Calbindin-positive cells in presubiculum form a structure of the presubiculum is most prominent in primates lattice surrounding acetylcholinesterase activity clusters, whereas Frontiers in Systems Neuroscience | www.frontiersin.org 14 April 2017 | Volume 11 | Article 20 and humans and was described in classical studies of cellular there are few calbindin-positive cells in superficial layers of architecture (Rose, 1927; Altschul, 1933; Braak, 1978). More retrosplenial granular cortex.

Frontiers in Systems Neuroscience | www.frontiersin.org 14 April 2017 | Volume 11 | Article 20 Killian et al. Page 7

Killian et al. Page 7 NIH-PA Author Manuscript NIH-PA Author Manuscript

Ebbesen (2017) General Discussion

of medial entorhinal cortex, Chapter 6). Elucidating if and how the Killian et al. elaborate modular cytoarchitechtonics contribute to the generation Page 7 of spatial responses in the parahippocampal region would be a major advance in our understanding of how anatomical microcircuits contribute to cortical computation and temporal coding in general.

NIH-PA Author Manuscript The striking spatial representations in the parahippocampal cortex Killian et al. Page 7 present a rare opportunity to tie single cell cortical computation to the ‘high-level’ cognitive tasks, which the grid cell system is solving. As mentioned in the introduction, the fact that plotting the two- dimensional spike patterns of parahippocampal neurons produces enormously elegant patterns lets it appear likely that these neurons NIH-PA Author Manuscript are involved in a cognitive task, which requires an ‘encoding’ of the environment. Interestingly, it remains an open question what these

cognitive tasks might be. Early interpretations suggestedNIH-PA Author Manuscript that the grid cell system might be primarily for navigation, contributing to path integration of self-motion (Bush et al., 2015; McNaughton et al., 2006; Moser and Moser, 2008; Rowland et al., 2016). How- Figure 13: Grid-cell-like responses of entorhinal neurons ever, grid cells have several dynamic features, such as non-uniform to saccades across a visual scene. expansion in novel environments (Barry et al., 2012; Hägglund Top: Example 10-second saccade paths across a visual stimulus. et al., 2016), drift (Hardcastle et al., 2015) and distortions of the Middle: Plots of eye position hexagonality (Krupic et al., 2015; Stensola et al., 2015). These dy- (gray) and spikes (red) reveal a grid-cell-like firing pattern namic features make grid cells less than ideal as the matric basis (spikes only plotted at loci of Figure 1. Spatial representation in the primate entorhinal cortex

NIH-PA Author Manuscript of an internal navigation system, since any read-out mechanism above-average firing rates). Bot- tom: Spatial firing rate maps a. Recordings were performed using a linear electrode array placed in the entorhinal cortex would need to track and correct for these dynamic effects (but on across visual scenes show mul- (red arrow). Three example 10-second scan paths are shown in yellow. b. An example of an the other hand, mammals also generally perform poorly in path tiple distinct firing fields (dva = degrees of visual angle). (Adapt- integration tasks (Etienne and Jeffery, 2004; Etienne et al., 1996)). entorhinal grid cell. Left: plots of eye position (gray) and spikes (red) reveal non-uniform ed from Killian et al., 2012, Further, several studies have implicated neurons in superficial me- with permission, Nature NPG) spatial density of spiking. For clarity, only spikes corresponding to locations of firing rate NIH-PA Author Manuscript dial entorhinal cortex in distinctly ‘non-navigation’ tasks, such as above half of the mean rate were plotted. Monkey name and unit number are indicated by memory consolidation (Kitamura et al., 2014, 2017), integration of sensory stimuli (timing lever the characters at the top. Middle: spatial firing rate maps show multiple distinct firing fields. presses in response to auditory cues, Aronov et al., 2017) and mapping of a visual scene (Killian et al., 2012, 2015). Intriguingly, in the latter case, the discharge pattern in response to saccades across the The maximum of the rate map (red) is given at the top. Right: the spatial periodicity of the

NIH-PA Author Manuscript visual scene was both grid-cell-like and border-cell-like (Figure 13), suggesting that perhaps the spatial firing fields is seen with spatial autocorrelations. The color scale limits are ±1 (blue-red), the Figure 1. Spatial representation in171 the primatemaximum entorhinal correlation cortexmagnitude, with green being 0 correlation. g = gridness score, dva = a. Recordings were performed using adegrees linear ofelectrode visual angle. array c. Percentages placed in theof cells entorhinal in the EC cortex and HPC with a significantly high

Figure 1. Spatial(red representation arrow).NIH-PA Author Manuscript Three in the example primate entorhinal10-second cortexgridness scan paths score are(G), shownor border in score yellow. (B). Theb. An black example line shows of thean 5% chance level, the a. Recordingsentorhinal were performed grid cell.using Left:a linear plots electrode of eye array dashedposition placed lines in(gray) representthe entorhinal and the spikes 95% cortex confidence(red) reveal level; non-uniform *P < 0.05, **P < 0.01. d. Grid cell spacing (red arrow). spatialThree example density 10-second of spiking. scan Forpaths clarity, are shown increasedonly in yellow.spikes with b. correspondingdistance An example from theof anto rhinal locations sulcus, ofconsistent firing ratewith a dorsal-ventral gradient in entorhinal grid cell. Left: plots of eye position (gray) and rodentsspikes (red)and bats. reveal Open non-uniform and closed circles identify the grid cells from each of the two Figure 1. Spatialspatial representation densityabove inof the spiking. halfprimate of For entorhinalthe clarity, mean onlycortex rate spikes were corresponding plotted.monkeys. Monkey to Right:locations name autocorrelations of andfiring unit rate number for representative are indicated grid cells by recorded at different locations a. Recordings were performed using a linear electrode array placed in the entorhinal cortex above half ofthe the characters mean rate were at the plotted. top. MonkeyMiddle: name spatial andmedial unitfiring to number the rate rhinal mapsare indicatedsulcus. show e. byGridnessmultiple and distinct border firingscores fields.are plotted for all cells recorded in (red arrow). Three example 10-second scan paths are shown in yellow. b. An example of an the charactersThe at the maximum top. Middle: of spatialthe rate firing map rate (red) maps is showthe given posterior multiple at the EC distinct top. (n = Right:193; firing red: fields.the cells spatial with significantperiodicity gridness of the scores, n = 23; blue: cells with entorhinal grid cell. Left: plots of eye position (gray) and spikes (red) reveal non-uniform The maximum of the rate map (red) is given at the top. Right:significant the spatial border periodicity scores, n of= 18).the spatial density of spiking. Forfiring clarity, fields only spikes is seen corresponding with spatial to locationsautocorrelations. of firing rate The color scale limits are ±1 (blue-red), the firing fields is seen with spatial autocorrelations. The color scale limits are ±1 (blue-red), the above half of the mean rate weremaximum plotted. Monkeycorrelation name magnitude,and unit number with are greenindicated being by 0 correlation. g = gridness score, dva = maximum correlation magnitude, with green being 0 correlation. g = gridness score, dva = the characters at the top. Middle:degrees spatial of firing visual rate angle. maps show c. Percentages multiple distinct of firingcells fields. in the EC and HPC with a significantly high The maximum ofdegrees the rate of map visual (red) angle. is given c. Percentages at the top. Right: of cells the in spatial the EC periodicity and HPC of with the a significantly high

NIH-PA Author Manuscript gridness score (G), or border score (B). The black line shows the 5% chance level, the NIH-PA Author Manuscript firing fields is seengridness with spatialscore (G),autocorrelations. or border score The (B). color The scale black limits line are shows ±1 (blue-red), the 5% chance the level, the maximum correlationdashed magnitude, lines dashedrepresent with lines greenthe 95% representbeing confidence 0 correlation. the level;95% g *=confidenceP gridness < 0.05, score,**P level; < dva0.01. =* d.P

NIH-PA Author Manuscript gridness score (G),rodents or border and bats.rodents score Open (B). and Theand bats. blackclosed Openline circles shows and identify the closed 5% the chance circlesgrid cellslevel, identify from the each the of gridthe two cells from each of the two dashed lines representmonkeys. the 95%Right:monkeys. confidence autocorrelations Right: level; *autocorrelations Pfor < 0.05,representative **P < 0.01. gridfor d. cellsrepresentativeGrid recordedcell spacing at differentgrid cells locations recorded at different locations increased with distancemedial tofrom the the rhinal rhinal sulcus. sulcus, e. Gridnessconsistent and with border a dorsal-ventral scores are gradient plotted forin all cells recorded in rodents and bats.the Open posterior and closedmedial EC ( ncircles = to193; the identify red: rhinal cells the sulcus.withgrid cellssignificant e. from Gridness each gridness of theand scores, two borderNature n .= Author scores23; blue: manuscript; are cells plotted withavailable for in PMC all 2013cells May recorded 29. in monkeys. Right:significant autocorrelations borderthe posteriorfor scores, representative n =EC 18). (n grid = 193; cells recordedred: cells at differentwith significant locations gridness scores, n = 23; blue: cells with medial to the rhinal sulcus. e.significant Gridness and border border scoresscores, are n plotted = 18). for all cells recorded in the posterior EC (n = 193; red: cells with significant gridness scores, n = 23; blue: cells with significant border scores, n = 18).

Nature. Author manuscript; available in PMC 2013 May 29.

Nature. Author manuscript; available inNature PMC 2013. Author May 29.manuscript; available in PMC 2013 May 29. Ebbesen (2017) Chapter 9 fields are not strictly tied to allocentric space, but rather the signature of an abstract, general principle by which cortical neurons encode complex information (Kerdels, 2016; Kerdels and Peters, 2015). Answering these many open questions (Figure 14) and determining what the grid cell system is really ‘doing’ will be a major advance in our understanding of the cellular basis of cognition.

9.2 Motor cortex and neural circuits underlying social exploration 9.2.1 What is the function of rat vibrissa motor cortex? The brain structure, which we refer to as ‘vibrissa motor cortex‘ in Chapters 7-8, is anatomically delineated as an agranular area with a large layer 5b and an (almost) absence of layer 4 (Brecht et al., 2004; Yamawaki et al., 2014; Zilles and Wree, 1995). Classical cortical mapping studies clearly assign this area as the primary motor representation of the whiskers (Brecht et al., 2004; Hall and Lindholm, 1974; Matyas et al., 2010; Neafsey et al., 1986) and accordingly many studies have focused on the relation between neural activity in this area and the kinematics of whisker movements (Gerdjikov et al., 2013; Hill et al., 2011; Sreenivasan et al., 2016). However – as we discuss in Chapter 8 – this brain structure is referred to by many names (Brecht, 2011) and has been implicated in a range of functions. For example, some studies have investigated this cortical area as a frontal orientation field (i.e. a rodent homologue of primate frontal eye field) and implicated the area in choice, working memory and preparation of upcoming motor decisions (Brody and Hanks, 2016; Erlich et al., 2011, 2015; Hanks et al., 2015). Other studies have investigated this area as secondary motor cortex (i.e. a homologue of primate supplementary motor areas) and implicated the neurons in decision making in relation to rewards and upcoming motor strategies (Barthas and Kwan, 2017; Murakami et al., 2014, 2016, Reep et al., 1987, 1990; Sul et al., 2011).

In the second part of this thesis I presented a study where we investigated the activity patterns in vibrissa motor cortex during a range of ‘naturalistic’ whisking behaviors (Chapter 7). We can draw two major conclusions from our observations. The first major conclusion is that spike patterns, microstimulation and inactivation experiments point to a major role of this cortical structure in inhibitory control of behavior. In the manuscript presented in Chapter 8, we already discussed in detail how this observation aligns with other studies on motor cortex function across mammals. The second major conclusion is that neural activity in primary motor cortex is modulated by social touch, even when basic kinematic aspects of the whisker movement are regressed out (at least in our first-order generalized

172 Ebbesen (2017) General Discussion linear modeling of the spike patterns). This point was not developed extensively in the manuscripts presented in Chapter 7-8, but it opens avenues for interesting follow-up studies investigating social computations in frontal cortex.

9.2.2 Social computations in sensorimotor cortex Social interactions, such as play behavior and pair bond formation, are fundamental aspects of animal and human behavior. Despite this fact, we still know only very little about the cortical machinery for social cognition. The somatosensory cortex is a large cortical area, which contains a remarkably detailed, somatotopic representation of the body (Lenschow et al., 2016; Woolsey and Van der Loos, 1970) and is massively interconnected with motor cortex (Barth et al., 1990; Diamond et al., 2008; Feldmeyer et al., 2013; Hatsopoulos and Suminski, 2011; Smith and Alloway, 2013). It has long been known that neural activity in soma- Whisker kinematics? tosensory cortex is criti- Orientation? Social computations? Memory consolidation? cal for the perception of Behavioral inhibition? Navigation? Integrating Ramp-to-threshold sensory touch, i.e. sensing the decisions? stimuli? body surface (Mount- Visual space? castle, 1956). Surprisingly, recent Generalized higher-order evidence shows that activity in soma- processing? tosensory cortex is also key for social cognition, i.e. ‘thinking about’ your own body and other bodies. For Figure 14: What are cortical neurons example, in rodents (Bobrov et al., 2014) and humans (Gaz- doing? Vibrissa motor cortex (red color) zola et al., 2012), neural activity in somatosensory cortex and parahippocampal cortex (purple color) has been implicated in a diverse is different when touching male and a female conspecifics, range of functions across a variety of even though the actual touch input is the same – a signa- cognitive domains. (Adapted from Poucet & Sargolini, 2013, with permis- ture that these neurons encode ’social’ categories. Similarly, sion, Nature Pub. Group) somatosensory cortex is activated in a social context before any actual touch input (Lenschow and Brecht, 2015), when observing others in pain (Keysers et al., 2010) , and when simply imagining pleasant or sexual touch (Wise et al., 2016), so-called ‘vicarious’ touch responses. Human studies using trans-cranial magnetic stimulation to influence cortical activity have shown that these ‘vicarious’ responses in somatosensory cortex are actually causally involved in crucial aspects of healthy social cognition, such as empathy (Bolognini et al., 2013; Keysers et al., 2010) and deciphering the bodily and emotional states of conspecifics (Paracampo et al., 2016; Val- chev et al., 2016). 173 Ebbesen (2017) Chapter 9

To investigate if motor cortical activity is modulated by social covariates further studies might record vibrissa motor cortex of both male and female animals, across hormonal states, during social facial interactions with both male and female conspecifics to determine if performing the same whisker movements in different social contexts modulates neural activity differentially. While such a study is beyond the scope of this thesis, I already observed response patterns during my experiments presented in Chapter 7, which indicated that motor cortical activity is modulated by social covariates. For example, even though whisking patterns in male rats are very similar between interactions with male and female conspecifics (Wolfe et al., 2011), I recorded neurons in the vibrissa motor cortex of male rats, which showed markedly different responses to interacting with male and female conspecifics. We are currently investigating these observations further and comparing them to social signals in sensory cortices (Bobrov et al., 2014; Rao et al., 2014).

It is presently unknown how higher-order ‘social’ responses in somatosensory cortex are generated from afferent sensory information relayed to cortex from the sensory epithelium. We also do not know if these responses in social contexts are truly ‘social’ in the sense that they are specific to the social domain (as e.g. the macaque face patch systems seems specifically tuned to recognize conspecifics (Ghazanfar and Santos, 2004; Tsao and Livingstone, 2008; Tsao et al., 2006)), or if they are simply ‘or- dinary’ responses to highly salient stimuli. Elucidating if and how activity patterns indicative of social computations (e.g. different responses to the same ‘touch’ from a male and female conspecific (Bobrov et al., 2014; Gazzola et al., 2012), modulation of these responses by the hormonal state of the animal (Bobrov et al., 2014)) extend to motor cortex could provide valuable information about how ‘social’ patterns of computation arise, how they propagate across neural circuits and what they contribute to the neural control of healthy social behavior.

9.3 References Alonso, A., and Klink, R. (1993). Differential electroresponsiveness of stellate and pyramidal-like cells of medial entorhinal cortex layer II. J. Neurophysiol. 70, 128–143. Aronov, D., Nevers, R., and Tank, D.W. (2017). Mapping of a non-spatial dimension by the hippocampal–entorhinal circuit. Nature 543, 719–722. Barry, C., Ginzberg, L.L., O’Keefe, J., and Burgess, N. (2012). Grid cell firing patterns signal environmental novelty by expansion. Proc. Natl. Acad. Sci. U. S. A. 109, 17687–17692.

174 Ebbesen (2017) General Discussion

Barth, T.M., Jones, T. a., and Schallert, T. (1990). Functional subdivisions of the rat somatic sensorimotor cortex. Behav. Brain Res. 39, 73–95. Barthas, F., and Kwan, A.C. (2017). Secondary Motor Cortex : Where “ Sensory ” Meets “ Motor ” in the Rodent Frontal Cortex. Trends Neurosci. 40, 181–193. Bobrov, E., Wolfe, J., Rao, R.P., and Brecht, M. (2014). The representation of social facial touch in rat barrel cortex. Curr. Biol. 24, 109–115. Boccara, C.N., Sargolini, F., Thoresen, V.H., Solstad, T., Witter, M.P., Moser, E.I., and Moser, M.-B. (2010). Grid cells in pre- and parasubiculum. Nat. Neurosci. 13, 987–994. Bolognini, N., Rossetti, a, Convento, S., and Vallar, G. (2013). Understanding other’s fellings: the role of the right primary somatosensory cortwx in encoding the affective valence of other’s touch. J. Neurosci. 33, 4201–4205. Brecht, M. (2011). Movement, Confusion, and Orienting in Frontal Cortices. Neuron 72, 193–196. Brecht, M., Krauss, A., Muhammad, S., Sinai-Esfahani, L., Bellanca, S., and Margrie, T.W. (2004). Organization of rat vibrissa motor cortex and adjacent areas according to cytoarchitectonics, microstimulation, and intracellular stimulation of identified cells. J. Comp. Neurol. 479, 360–373. Brecht, M., Ray, S., Burgalossi, A., Tang, Q., Schmidt, H., and Naumann, R. (2014). An isomorphic mapping hypothesis of the grid representation. Philos. Trans. R. Soc. Lond. B. Biol. Sci. 369, 20120521. Brody, C.D., and Hanks, T.D. (2016). Neural underpinnings of the evidence accumulator. Curr. Opin. Neurobiol. 37, 149–157. Burgalossi, A., and Brecht, M. (2014). Cellular, columnar and modular organization of spatial representations in medial entorhinal cortex. Curr. Opin. Neurobiol. 24, 47–54. Burgalossi, A., Herfst, L., von Heimendahl, M., Förste, H., Haskic, K., Schmidt, M., and Brecht, M. (2011). Microcircuits of Functionally Identified Neurons in the Rat Medial Entorhinal Cortex. Neuron 70, 773–786. Bush, D., Barry, C., Manson, D., Burgess, N., Blair, H.T., Save, E., Moser, E.I., Merkow, M., Polyn, S.M., Jacobs, J., et al. (2015). Using Grid Cells for Navigation. Neuron 87, 507–520. Diamond, M.E., von Heimendahl, M., Knutsen, P.M., Kleinfeld, D., and Ahissar, E. (2008). “Where” and “what” in the whisker sensorimotor system. Nat. Rev. Neurosci. 9, 601–612. Domnisoru, C., Kinkhabwala, A.A., and Tank, D.W. (2013). Membrane potential dynamics of grid cells. Nature 495, 199–204. Erlich, J.C., Bialek, M., and Brody, C.D. (2011). A Cortical Substrate for Memory-Guided Orienting in the Rat. Neuron 72, 330–343. 175 Ebbesen (2017) Chapter 9

Erlich, J.C., Brunton, B.W., Duan, C.A., Hanks, T.D., and Brody, C.D. (2015). Distinct effects of prefrontal and parietal cortex inactivations on an accumulation of evidence task in the rat. 1–28. Etienne, A.S., and Jeffery, K.J. (2004). Path integration in mammals. Hippocampus 14, 180–192. Etienne, A.S., Maurer, R., and Séguinot, V. (1996). Path integration in mammals and its interaction with visual landmarks. J. Exp. Biol. 199, 201–209. Feldmeyer, D., Brecht, M., Helmchen, F., Petersen, C.C.H., Poulet, J.F.A., Staiger, J.F., Luhmann, H.J., and Schwarz, C. (2013). Barrel cortex function. Prog. Neurobiol. 103, 3–27. Fuchs, E.C., Neitz, A., Pinna, R., Melzer, S., Caputi, A., and Monyer, H. (2016). Local and Distant Input Controlling Excitation in Layer II of the Medial Entorhinal Cortex. Neuron 89, 194–208. Gazzola, V., Spezio, M.L., Etzel, J. a., Castelli, F., Adolphs, R., and Keysers, C. (2012). PNAS Plus: Primary somatosensory cortex discriminates affective significance in social touch. Proc. Natl. Acad. Sci. 109, E1657–E1666. Gerdjikov, T. V, Haiss, F., Rodriguez-Sierra, O.E., and Schwarz, C. (2013). Rhythmic whisking area (RW) in rat primary motor cortex: an internal monitor of movement-related signals? J. Neuro- sci. 33, 14193–14204. Ghazanfar, A.A., and Santos, L.R. (2004). Primate brains in the wild: the sensory bases for social interactions. Nat. Rev. Neurosci. 5, 603–616. Giocomo, L.M., Moser, M.-B., and Moser, E.I. (2011). Computational models of grid cells. Neuron 71, 589–603. Hall, R.D., and Lindholm, E.P. (1974). Organization of motor and somatosensory neocortex in the albino rat. Brain Res. 66, 23–38. Hanks, T.D., Kopec, C.D., Brunton, B.W., Duan, C.A., Erlich, J.C., and Brody, C.D. (2015). Dis- tinct relationships of parietal and prefrontal cortices to evidence accumulation. Nature 520, 220–223. Hardcastle, K., Ganguli, S., and Giocomo, L.M. (2015). Environmental Boundaries as an Error Cor- rection Mechanism for Grid Cells. Neuron 86, 827–839. Hatsopoulos, N.G., and Suminski, A.J. (2011). Sensing with the motor cortex. Neuron 72, 477–487. Henn-Mike, N., Kecskés, M., Krabóth, Z., Sebestyén, H., and Varga, C. (2016). KKK10 183.17 Me- dial entorhinal cortex layer III pyramidal cells form dendritic bundles in a grid-like hexagonal pattern and receive specific inputs from the presubiculum. In Neuroscience 2016, (San Diego), Hill, D.N., Curtis, J.C., Moore, J.D., and Kleinfeld, D. (2011). Primary motor cortex reports efferent control of vibrissa motion on multiple timescales. Neuron 72, 344–356. Hägglund, M., Moser, M.-B., and Moser, E.I. (2016). 183.06/JJJ60 Spatial dynamics of grid cells dur- 176 Ebbesen (2017) General Discussion

ing novelty. In Neuroscience 2016, (San Diego). Kerdels, J. (2016). A Computational Model of Grid Cells based on a Recursive Growing Neural Gas. Kerdels, J., and Peters, G. (2015). A New View on Grid Cells Beyond the Cognitive Map Hypothesis. 1–10. Keysers, C., Kaas, J.H., and Gazzola, V. (2010). Somatosensation in social perception. Nat. Rev. Neu- rosci. 11, 417–428. Killian, N.J., Jutras, M.J., and Buffalo, E. a. (2012). A map of visual space in the primate entorhinal cortex. Nature 491(7426): 761–764. Killian, N.J., Potter, S.M., and Buffalo, E.A. (2015). Saccade direction encoding in the primate entorhinal cortex during visual exploration. Proc. Natl. Acad. Sci. 112(51):15743-8 Kitamura, T., Pignatelli, M., Suh, J., Kohara, K., Yoshiki, A., Abe, K., and Tonegawa, S. (2014). Island cells control temporal association memory. Science 343, 896–901. Kitamura, T., Ogawa, S.K., Roy, D.S., Okuyama, T., Morrissey, M.D., Smith, L.M., Redondo, R.L., and Tonegawa, S. (2017). Engrams and circuits crucial for systems consolidation of a memory. Science (80-. ). 356. Krupic, J., Bauza, M., Burton, S., Barry, C., and O’Keefe, J. (2015). Grid cell symmetry is shaped by environmental geometry. Nature 518, 232–235. Latuske, P., Toader, O., and Allen, K. (2015). Interspike Intervals Reveal Functionally Distinct Cell Populations in the Medial Entorhinal Cortex. J. Neurosci. 35, 10963–10976. Lenschow, C., and Brecht, M. (2015). Barrel Cortex Membrane Potential Dynamics in Social Touch. Neuron 85, 718–725. Lenschow, C., Copley, S., Gardiner, J.M., Talbot, Z.N., Vitenzon, A., and Brecht, M. (2016). Sexually Monomorphic Maps and Dimorphic Responses in Rat Genital Cortex. Curr. Biol. 26, 106–113. Matyas, F., Sreenivasan, V., Marbach, F., Wacongne, C., Barsy, B., Mateo, C., Aronoff, R., and Pe- tersen, C.C.H. (2010). Motor control by sensory cortex. Science 330, 1240–1243. McNaughton, B.L., Battaglia, F.P., Jensen, O., Moser, E.I., and Moser, M.-B. (2006). Path integration and the neural basis of the “cognitive map.” Nat. Rev. Neurosci. 7, 663–678. Moser, E.I., and Moser, M.-B. (2008). A metric for space. Hippocampus 18, 1142–1156. Moser, E.I., Kropff, E., and Moser, M.-B. (2008). Place Cells, Grid Cells, and the Brain’s Spatial Rep- resentation System. Annu. Rev. Neurosci. 31, 69–89. Mountcastle, V. (1956). Modalily and topographic properties of single neurons of cat’s somatic sensory system. J.Neurophysiol. 20, 408–434. Muñoz, W., Tremblay, R., Rudy, B., Rudy, B., Silberberg, G., Luo, J., Markram, H., Fu, Y., Lu, J., Lin, 177 Ebbesen (2017) Chapter 9

Y., et al. (2014). Channelrhodopsin-Assisted Patching: In Vivo Recording of Genetically and Morphologically Identified Neurons throughout the Brain. Cell Rep. 9, 2304–2316. Murakami, M., Vicente, M.I., Costa, G.M., and Mainen, Z.F. (2014). Neural antecedents of self- initiated actions in secondary motor cortex. Nat. Neurosci. 17, 1574–1582. Murakami, M., Shteingart, H., Loewenstein, Y., and Mainen, Z.F. (2016). Distinct sources of deterministic and stochastic components of action timing decisions in rodent frontal cortex. bioRxiv. Naumann, R.K., Ray, S., Prokop, S., Las, L., Heppner, F.L., and Brecht, M. (2016). Conserved size and periodicity of pyramidal patches in layer 2 of medial/caudal entorhinal cortex. J. Comp. Neurol. 524, 783–806. Neafsey, E.J., Bold, E.L., Haas, G., Hurley-Gius, K.M., Quirk, G., Sievert, C.F., and Terreberry, R.R. (1986). The organization of the rat motor cortex: A microstimulation mapping study. Brain Res. Rev. 11, 77–96. Newman, E.L., and Hasselmo, M.E. (2014). Grid cell firing properties vary as a function of theta phase locking preferences in the rat medial entorhinal cortex. Front. Syst. Neurosci. 8, 1–15. Paracampo, R., Tidoni, E., Borgomaneri, S., di Pellegrino, G., and Avenanti, A. (2016). Sensorimotor Network Crucial for Inferring Amusement from Smiles. Cereb. Cortex 1–14. Preston-Ferrer, P., Coletta, S., Frey, M., and Burgalossi, A. (2016). Anatomical organization of presubicular head-direction circuits. Elife 5. Rao, R.P., Mielke, F., Bobrov, E., and Brecht, M. (2014). Vocalization–whisking coordination and multisensory integration of social signals in rat auditory cortex. Elife 3, 1–20. Ray, S., Naumann, R., Burgalossi, A., Tang, Q., Schmidt, H., and Brecht, M. (2014). Grid-layout and theta-modulation of layer 2 pyramidal neurons in medial entorhinal cortex. Science 343, 891–896. Ray, S., Burgalossi, A., Brecht, M., and Naumann, R.K. (2017). Complementary Modular Microcir- cuits of the Rat Medial Entorhinal Cortex. Front. Syst. Neurosci. 11, 1–18. Reep, R.L., Corwin, J. V, Hashimoto, A., and Watson, R.T. (1987). Efferent connections of the rostral portion of medial agranular cortex in rats. Brain Res. Bull. 19, 203–221. Reep, R.L., Goodwin, G.S., and Corwin, J. V. (1990). Topographic organization in the corticocortical connections of medial agranular cortex in rats. J. Comp. Neurol. 294, 262–280. Reifenstein, E.T., Ebbesen, C.L., Tang, Q., Brecht, M., Schreiber, S., and Kempter, R. (2016). Cell-Type Specific Phase Precession in Layer II of the Medial Entorhinal Cortex. J. Neurosci. 36(7):2283-8. Rowland, D.C., Roudi, Y., Moser, M.-B., and Moser, E.I. (2016). Ten Years of Grid Cells. Annu. Rev. 178 Ebbesen (2017) General Discussion

Neurosci. 39, 19–40. Savelli, F., and Knierim, J.J. (2014). Strides toward a structure-function understanding of cortical representations of allocentric space. Neuron 84, 1108–1109. Shay, C.F., Ferrante, M., Chapman, G.W., and Hasselmo, M.E. (2015). Rebound spiking in layer II medial entorhinal cortex stellate cells: Possible mechanism of grid cell function. Neurobiol. Learn. Mem. Smith, J.B., and Alloway, K.D. (2013). Rat whisker motor cortex is subdivided into sensory-input and motor-output areas. Front. Neural Circuits 7, 4. Sreenivasan, V., Esmaeili, V., Kiritani, T., Galan, K., Crochet, S., and Petersen, C.C.H. (2016). Move- ment Initiation Signals in Mouse Whisker Motor Cortex. Neuron 92, 1368–1382. Stensola, H., Stensola, T., Solstad, T., Frøland, K., Moser, M.-B., and Moser, E.I. (2012). The entorhinal grid map is discretized. Nature 492, 72–78. Stensola, T., Stensola, H., Moser, M.-B., and Moser, E.I. (2015). Shearing-induced asymmetry in entorhinal grid cells. Nature 518, 207–212. Sul, J.H., Jo, S., Lee, D., and Jung, M.W. (2011). Role of rodent secondary motor cortex in value- based action selection. Nat. Neurosci. 14, 1202–1208. Sun, C., Kitamura, T., Yamamoto, J., Martin, J., Pignatelli, M., Kitch, L.J., Schnitzer, M.J., and Tonegawa, S. (2015). Distinct speed dependence of entorhinal island and ocean cells, including respective grid cells. Proc. Natl. Acad. Sci. 112, 9466–9471. Tang, Q., Ebbesen, C.L., Sanguinetti-scheck, J.I., Preston-ferrer, P., Gundlfinger, A., Winterer, J., Beed, P., Ray, S., Naumann, R., Schmitz, D., et al. (2015). Anatomical Organization and Spati- otemporal Firing Patterns of Layer 3 Neurons in the Rat Medial Entorhinal Cortex. J. Neurosci. 35, 12346–12354. Toader, O. (2016). Genetic approaches to probe spatial coding in the medial entorhinal cortex. Ruper- to-Carola University of Heidelberg, Germany. Tsao, D.Y., and Livingstone, M.S. (2008). Mechanisms of face perception. Annu. Rev. Neurosci. 31, 411–437. Tsao, D.Y., Freiwald, W.A., Tootell, R.B.H., and Livingstone, M.S. (2006). A cortical region consist- ing entirely of face-selective cells. Science 311, 670–674. Valchev, N., Gazzola, V., Avenanti, A., and Keysers, C. (2016). Primary somatosensory contribution to action observation brain activity-combining fMRI and cTBS. Soc. Cogn. Affect. Neurosci. 11, 1205–1217. Wise, N.J., Frangos, E., and Komisaruk, B.R. (2016). Activation of sensory cortex by imagined genital 179 Ebbesen (2017) Chapter 9

stimulation: an fMRI analysis. Socioaffective Neurosci. Psychol. 6. Wolfe, J., Mende, C., and Brecht, M. (2011). Social facial touch in rats. Behav. Neurosci. 125, 900– 910. Woolsey, T.A., and Van der Loos, H. (1970). The structural organization of layer IV in the somatosensory region (S I) of mouse cerebral cortex: The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res. 17, 205–242. Yamawaki, N., Borges, K., Suter, B.A., Harris, K.D., Shepherd, G.M.G., Wood, L., Shepherd, G., Shepherd, G., Brockway, K., Byrnes, E., et al. (2014). A genuine layer 4 in motor cortex with prototypical synaptic circuit connectivity. Elife 4, e05422. Yoon, K., Lewallen, S., Kinkhabwala, A.A., Tank, D.W., and Fiete, I.R. (2016). Grid Cell Responses in 1D Environments Assessed as Slices through a 2D Lattice. Neuron 89, 1086–1099. Zilles, K., and Wree, A. (1995). Cortex: Areal and Laminar Structure. In The Rat Nevous System, G. Paxinos, ed. (San Diego: Academic Press), pp. 649–685. Zilli, E. a. (2012). Models of Grid Cell Spatial Firing Published 2005–2011. Front. Neural Circuits 6, 1–17.

180 Ebbesen (2017) General Discussion

181 Ebbesen (2017) Chapter 10 10. Curriculum vitae

- - - CV removed from published version - - -

182 Ebbesen (2017) Curriculum vitae

- - - CV removed from published version - - -

183 Ebbesen (2017) Chapter 11 11. List of publications Christian L. Ebbesen M.SC. · PH.D. STUDENT · NEUROSCIENCE XX Straße XX, XXXXX Berlin, Germany  +49 XXX XXXX XXXX |  [email protected] |  http://orcid.org/XXXX-XXXX-XXXX-XXXX

Publications Journal articles: Ebbesen, C.L., & Brecht, M. (2017). Motor cortex – To act or not to act? (submitted)

Ebbesen, C.L., Doron, G., Lenschow, C., Brecht, M. (2017). Vibrissa motor cortex activity suppresses contralateral whisking behavior. Nature Neu- roscience 20(1):82-89 Read online Previewed by: Kim, J. & Hires, A. (2017) Brake and gas pedals in motor cortex Nature Neuroscience, 20(1):4-6. Read online

Ebbesen,↑ C.L., Reifenstein, E.T., Tang, Q., Burgalossi, A., Ray, S., Schreiber, S., Kempter, R. & Brecht, M. (2016) Cell type-specific differences in spike timing and spike shape in rat parasubiculum and superficial medial entorhinal cortex. Cell Reports 16(4), pp. 1005-1015. Read online

(*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., (*)Sanguinetti-Scheck, J.I., Schmidt, H., Tukker, J.J., Naumann, R., Ray, S., Preston-Ferrer, P., Schmitz, D., Brecht, M. (2016) Functional Architecture of the Rat Parasubiculum. Journal of Neuroscience 36(7):2289-2301. (*) Equally contributing Read online ”Featured Article”, previewed by: Esch, T. (2016) This Week in The Journal: Anatomy and Physiology of Parasubiculum Journal of Neu- roscience, 36(7):i-i. Read online ↑ Reifenstein, E.T., Ebbesen, C.L., Tang, Q., Brecht, M., Schreiber, S., Kempter, R. (2016) Cell-Type Specific Phase Precession in Layer II of the Medial Entorhinal Cortex.Journal of Neuroscience 36(7):2283-8. Read online

Tang, Q., Ebbesen, C.L., Sanguinetti, J.I., Preston-Ferrer, P., Gundlfinger, A., Winterer, J., Beed, P., Ray, S., Naumann, R., Schmitz, D., Brecht, M., & Burgalossi, A. (2015) Anatomical organization and spatio-temporal firing properties of layer 3 neurons in the rat medial entorhinal cortex. Journal of Neuroscience, 35(36):12346-12354 Read online Published during PhD studies PhD during Published (*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Spicher, D. & Brecht, M. (2014) Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex. Neuron, 84(6):1191-1197. (*) Equally contributing Read online Previewed by: Savelli, F. & Knierim, J.J. (2014) Strides toward a Structure-Function Understanding of Cortical Representations of Allo- centric Space Neuron, 84(6):1108-1109. Read online ↑ Ebbesen, C.L. & Bruus, H. (2012) Analysis of laser-induced heating in optical neuronal guidance. Journal of Neuroscience Methods, 209(1):168- 177 Read online

(*)Adams, J.D., (*)Ebbesen, C.L., Barnkob, R., Yang, A.H.J. , Soh, H.T., & Bruus, H. (2012) High-throughput, temperature-controlled micro- channel acoustophoresis device made with rapid prototyping. Journal of Micromechanics and Microengineering. 22:075017 (*) Equally contributing Read online

Peer-reviewed conference papers: Ebbesen, C.L., Adams, J.D., Barnkob, R., Soh, H.T. & Bruus, H. (2010): Temperature-Controlled High-Throughput (1 L/h) Acoustophoretic Parti- cle Separation in Microchannels. Proceedings of 14th International Conference on Miniaturized Systems for Chemistry and Life Sciences (μTAS) 728-730 (peer reviewed conference paper) Read online

JUNE 1, 2018 CHRISTIAN L. EBBESEN · LIST OF PUBLICATIONS 1 184 Ebbesen (2017) Declaration of contribution 12. Declaration of contribution The German version shall prevail.

Anlage zu § 6, Abs. 2 der Promotionsordnung der Lebenswissenschaftlichen Fakultät vom 05.03.2015, veröffentlicht in: Amtliches Mitteilungsblatt HU 12/2015 Annex to § 6, para. 2 of the Doctoral Degree Regulations of the Faculty of Life Sciences, amended on 05.03.2015, published in: University Gazette of Humboldt-Universität zu Berlin 12/2015

Erklärung über den Eigenanteil an den veröffentlichten bzw. zur Veröffentlichung angenommenen wissenschaftlichen Schriften innerhalb meiner Dissertationsschrift gemäß § 6, Abs. 2 der Promotionsordnung Declaration regarding my own contribution to the published academic papers, or academic papers which have been accepted for publishing, within my doctoral thesis, under the provisions of § 6, para. 2 of the Doctoral Degree Regulations

Von der Antragstellerin/Vom Antragsteller einzutragen: | To be completed by the applicant:

1. Name, Vorname | Ebbesen, Christian Laut Institut (ggf. externe Einrichtung) | Bernstein Center for Computational Neuroscience Promotionsfach | Biologie Thema der Dissertation | Cortical circuits underlying social and spatial exploration in rats

2. Nummerierte Aufstellung der eingereichten Schriften (Titel, Autor/innen, wo und wann veröffentlicht bzw. eingereicht): Numbered breakdown of the submitted papers (title, authors, where and when published or submitted):

1. (*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., Ray, S., Naumann, R., Schmidt, H., Spicher, D. & Brecht, M. (2014) Pyramidal and Stellate Cell Specificity of Grid and Border Representations in Layer 2 of Medial Entorhinal Cortex. Neuron 84(6):1191- 1197.

2. (*)Tang, Q., (*)Burgalossi, A., (*)Ebbesen, C.L., (*)Sanguinetti-Scheck, J.I., Schmidt, H., Tukker, J.J., Naumann, R., Ray, S., Preston-Ferrer, P., Schmitz, D., Brecht, M. (2016) Functional Architecture of the Rat Parasubiculum. Journal of Neuroscience 36(7):2289-2301.

3. Ebbesen, C.L., Reifenstein, E.T., Tang, Q., Burgalossi, A., Ray, S., Schreiber, S., Kempter, R. & Brecht, M. (2016) Cell type-specific differences in spike timing and spike shape in rat parasubiculum and superficial medial entorhinal cortex. Cell Reports 16(4), pp. 1005-1015.

4. Ebbesen, C.L., Doron, G., Lenschow, C., Brecht, M. (2017). Vibrissa motor cortex activity suppresses contralateral whisking behavior. Nature Neuroscience 20(1):82- 89.

5. Ebbesen, C.L., & Brecht, M. (2017). Motor cortex – To act or not to act? (submitted to Nature Reviews Neuroscience on 21.03.2017)

(*) Equal contribution.

3. Darlegung des eigenen Anteils an diesen Schriften: Statement regarding my own contribution to these papers:

Erläuterung: Legen Sie dar, welche von Ihnen geleisteten Arbeiten diese Schriften enthalten (z. B. Entwicklung der Konzeption, Literaturrecherche, Methodenentwicklung,

1 185 Ebbesen (2017) Chapter 12

The German version shall prevail.

Versuchsdesign, Datenerhebung, Datenauswertung, Ergebnisdiskussion, Erstellen des Manuskripts, Programmierung, Beweisführung) und wie groß Ihr Anteil (z. B. vollständig, überwiegend, mehrheitlich, in Teilen) an den jeweiligen Leistungen war. Explanation: State which of the works which you rendered comprise these papers (e. g. development of the conception, literature research, development of methods, test design, data collection, data analysis, results discussion, compiling the manuscript, programming, reasoning), and how large your contribution (e. g. entire, large, predominant, partial) was the respective works.

For each publication included in the thesis, I provide a delineation of author contributions (as published, if available), followed by a rough assessment of my contribution as 0 (none), 1 (minor, ~0-25%), 2 (partial, ~25-50%), 3 (substantial, ~50- 75%), 4 (major, 75-100%).

zu Nr. 1 | to no. 1 Author contributions: Q.T. and A.B. performed juxtacellular recordings. C.L.E. and Q.T. performed tetrode recordings. S.R., R.N., and H.S. performed and analyzed anatomical experiments. C.L.E. and D.S. developed the classifier and C.L.E. and Q.T. analyzed electrophysiology data. A.B. and M.B. conceived of the project and supervised experiments. All authors contributed to the writing of the manuscript. Estimated contribution: Design of Experiments 1 Acquisition of Data 2 Analysis of Data 4 Writing of Manuscript 2

zu Nr. 2 | to no. 2 Author contributions: Q.T., A.B., C.L.E., J.I.S.-S., and M.B. designed research; Q.T., A.B., C.L.E., J.I.S.-S., H.S., J.J.T., R.N., S.R., P.P.-F., D.S., and M.B. performed research; Q.T., A.B., C.L.E., J.I.S.-S., H.S., J.J.T., R.N., S.R., P.P.-F., D.S., and M.B. analyzed data; Q.T., A.B., C.L.E., J.I.S.-S., H.S., J.J.T., R.N., S.R., P.P.-F., D.S., and M.B. wrote the paper. Design of Experiments 2 Acquisition of Data 2 Analysis of Data 4 Writing of Manuscript 2

zu Nr. 3 | to no. 3 Author contributions: C.L.E. analyzed burstiness, spike shape, theta rhythmicity, and cycle skipping. E.T.R. analyzed phase precession. Q.T. and A.B. provided access to juxtacellular data and assisted with data analysis. S.R. performed immunohistochemstry and microscopy. S.S., R.K., and M.B. provided expertise and feedback on the analysis and supervised the project. C.L.E. conceived the study and wrote the first version of the manuscript. All authors provided feedback and contributed to writing the manuscript. Design of Experiments 4 Acquisition of Data 0 Analysis of Data 4 Writing of Manuscript 4

zu Nr. 4 | to no. 4 Author contributions: C.L.E., G.D. and M.B. designed the study. C.L.E. performed tetrode experiments. C.L.E. and G.D. performed juxtacellular experiments. G.D. and C.L. performed whole-cell recordings. C.L.E. and G.D. performed microstimulation and blockade experiments. C.L.E. analyzed the data and performed statistical

2 186 Ebbesen (2017) Declaration of contribution

The German version shall prevail.

modeling.C.L.E. and M.B. wrote the first version of the manuscript. All authors assisted with analyzing data and contributed to writing the manuscript. Design of Experiments 4 Acquisition of Data 3 Analysis of Data 4 Writing of Manuscript 4

zu Nr. 5 | to no. 5 Author contributions: C.L.E. and M.B. designed the structure of the review. C.L.E. performed literature research, made figures and wrote the first version of the manuscript. M.B. provided feedback on figures. Both authors contributed to writing the manuscript. Literature research 4 Writing of Manuscript 4

4. Anschriften und E-Mail-Adressen der jeweiligen Mitautor/innen: Postal and e-mail addresses of the respective co-authors:

zu Nr. 1 | to no. 1 (*)Tang, Q.1, (*)Burgalossi, A.2, (*)Ebbesen, C.L.1, Ray, S.1, Naumann, R.3, Schmidt, H.1, Spicher, D.1 & Brecht, M.1 1Bernstein Center for Computational Neuroscience, Humboldt Universität zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany 2Werner Reichardt Centre for Integrative Neuroscience, Otfried-Müller-str. 25, 72076 Tübingen, Germany 3Max Planck Institute for Brain Research, Max-von-Laue-Str. 4, 60438 Frankfurt am Main, Germany E-mails: [email protected] (Q.T.), [email protected] tuebingen.de (A.B.), [email protected] (C.E.), [email protected] (S.R.), [email protected] (R.N.), [email protected] (H.S.), [email protected] (D.S.), [email protected] (M.B.)

zu Nr. 2 | to no. 2 (*)Tang, Q.1, (*)Burgalossi, A.2, (*)Ebbesen, C.L.1, (*)Sanguinetti-Scheck, J.I.1, Schmidt, H.1, Tukker, J.J.3, Naumann, R.4, Ray, S.1, Preston-Ferrer, P.2, Schmitz, D.3, Brecht, M.1 1Bernstein Center for Computational Neuroscience, Humboldt Universität zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany 2Werner Reichardt Centre for Integrative Neuroscience, Otfried-Müller-str. 25, 72076 Tübingen, Germany 3Charité Universitätsmedizin, Charitéplatz 1, Berlin, 10117 Berlin, Germany 4Max Planck Institute for Brain Research, Max-von-Laue-Str. 4, 60438 Frankfurt am Main, Germany E-mails: [email protected] (Q.T.), [email protected] tuebingen.de (A.B.), [email protected] (C.E.), juan.sanguinetti@bccn- berlin.de (J.S.), [email protected] (H.S.), [email protected] (J.T.), [email protected] (R.N.), [email protected] (S.R.), [email protected] (P.P.), [email protected] (D.S.), [email protected] (M.B.)

zu Nr. 3 | to no. 3 Ebbesen, C.L.1, Reifenstein, E.T.2, Tang, Q.1, Burgalossi, A.3, Ray, S.1, Schreiber, S.2, Kempter, R.2 & Brecht, M.1 1Bernstein Center for Computational Neuroscience, Humboldt Universität zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany

3 187 Ebbesen (2017) Chapter 12

The German version shall prevail.

2Institute for Theoretical Biology and Department of Biology, Humboldt Universität zu Berlin, 10115 Berlin, Germany 3Werner Reichardt Centre for Integrative Neuroscience, Otfried-Müller-str. 25, 72076 Tübingen, Germany E-mails: [email protected] (C.E.), [email protected] (E.R.), [email protected] (Q.T.), [email protected] (A.B.), [email protected] (S.R.), [email protected] (S.S.), [email protected] (R.K.), [email protected] (M.B.)

zu Nr. 4 | to no. 4 Ebbesen, C.L.1, Doron, G.2, Lenschow, C.1 & Brecht, M.1 1Bernstein Center for Computational Neuroscience, Humboldt Universität zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany 2Charité Universitätsmedizin, Charitéplatz 1, Berlin, 10117 Berlin, Germany E-mails: [email protected] (C.E.), [email protected] (G.D.), [email protected] (C.L.), [email protected] (M.B.)

zu Nr. 5 | to no. 5 Ebbesen, C.L.1 & Brecht, M.1 1Bernstein Center for Computational Neuroscience, Humboldt Universität zu Berlin, Philippstr. 13, Haus 6, 10115 Berlin, Germany E-mails: [email protected] (C.E.), [email protected] (M.B.)

Datum, Unterschrift der Antragstellerin/des Antragstellers: Date, signature of the applicant:

Date: 01.04.2017

Die Angaben zu Punkt 3 müssen von den Mitautor/innen schriftlich bestätigt werden. The information submitted in point 3 must be confirmed in writing by the co-authors.

Ich bestätige die von Herrn Ebbesen unter Punkt 3 abgegebene Erklärung: I confirm the declaration submitted by Mr. Ebbesen in point 3:

Name: Dr. Qiusong Tang Unterschrift: |

Name: Dr. Andrea Burgalossi Unterschrift: |

Name: Dr. Saikat Ray Unterschrift: |

Name: Dr. Robert Naumann Unterschrift: |

Name: Helene Schmidt Unterschrift: |

Name: Dominik Spicher Unterschrift: |

Name: Prof. Dr. Michael Brecht Unterschrift: |

4 188 Ebbesen (2017) Declaration of contribution

The German version shall prevail.

Name: Juan-Ignacio Sanguinetti-Scheck Unterschrift: |

Name: John J. Tukker, DPhil Unterschrift: |

Name: Dr. Patricia Preston-Ferrer Unterschrift: |

Name: Prof. Dr. Dietmar Schmitz Unterschrift: |

Name: Dr. Eric Reifenstein Unterschrift: |

Name: Prof. Dr. Susanne Schreiber Unterschrift: |

Name: Prof. Dr. Richard Kempter Unterschrift: |

5 189 Ebbesen (2017) Chapter 13 13. Acknowledgements So many people have helped me during my PhD studies and contributed to making my time in the Brecht lab enjoyable, productive and educational.

I would like to thank my supervisor Prof. Dr. Michael Brecht for fun, creative and exciting neuroscience. I have learned an incredible amount from your mentoring and I am thankful for the opportunity to work with you.

I would like to thank my supervisor Prof. Dr. Gabriel Curio for always providing thoughtful and in- sightful observations and discussions. Your perspectives have been educational and very helpful.

I would like to thank everyone at the Bernstein Center for their exceptional help, particularly Brigitte Geue, Undine Schneeweiß and Juliane Steger for technical assistance in the lab, Maik Kunert and Ar- nold Stern for technical support, Willi Schiegel and Rike-Benjamin Schuppner for systems managing, Andreea Neukirchner and Margret Franke for administrative help, and Susanne Grüßel and Denise Kaufmann for animal caretaking.

I would like to thank everyone at the Berlin School of Mind and Brain, especially Dr. Dirk Mende and Dr. Inken Dose for outstanding support and guidance.

I would like to thank everyone who warmly welcomed me into the Brecht lab (Dr. Qiusong Tang, Dr. Rajnish Rao, Dr. Saikat Ray, Dr. Edith Chorev, Helene Schmidt, Dr. Andrea Burgalossi, Dr. Patricia Preston Ferrer, Dr. Evgeny Bobrov, Dr. Guy Doron, John J. Tukker, DPhil, Dr. Constanze Lenschow, Dr. Moritz von Heimendahl, Dr. Robert Naumann, Falk Mielke and Viktor Bahr) and to my current colleagues (Peter Bennett, Dr. Ann Clemens, Konstantin Hartmann, Dr. Shimpei Ishiyama, Eduard Maier, Nacho Sanguinetti-Scheck, Johanna Sigl-Glöckner, Dr. Jean Simonnet and Simon Lauer). You helped me enormously, made my time in the lab fun, exciting and unforgettable. I am deeply grateful to all of you.

I would like to thank all other scientific collaborators, especially Dr. Eric Reifenstein, Prof. Dr. Su- sanne Schreiber, Prof. Dr. Richard Kempter, Dominik Spicher, Dr. Anja Gundlfinger, Dr. Jochen Winterer, Dr. Prateep Beed and Prof. Dr. Dietmar Schmitz.

190 Ebbesen (2017) Acknowledgements

I would like to thank everyone in the vibrant Berlin neuroscience community, especially the Jamichew Social (both founding members and all subsequently associated groups).

I would like to thank Dr. Mikkel Vestergaard and Dr. Ryszard Auksztulewicz for invaluable help and guidance, scientifically and otherwise.

I would like to thank my parents Ingrid and Jørgen, my sister Eva Maria, and all of my friends and family for all your support and encouragement.

Finally, I would like to thank Sara Helgheim Tawfiq for numerous 2d and 3d drawings for my papers, for incredible self-restraint when witnessing how I (ab)use Adobe Illustrator, for endless patience and understanding during periods of lab-related stress and for being the most special lady in my life.

191